Tag: TG_L2

Page Rank. The factor will be remarked.

This is the news (determined by the characteristic ((http://wiki.yandex-team.ru/jandekspoisk/kachestvopoiska/obshayaformula/tekushichiekomponenty/klassificacionnye?v=tkd#h45859-3 Patterns in URL URL)))).

This is a catalog (determined by the characteristic ((http://wiki.yandex-team.ru/jandekspoisk/kachestvopoiska/obshayafformula/tekushhiekomponenty/klassificacionnye? .

Attendance from the bar - ((http://wiki.yandex-team.ru/andrejjkostjagin/yabarlog/hoststat data description)). The factor will be remarked.

Long document (the longer the document, the greater the value of the factor).

The amount of IDF words of the request. The name does not reflect the essence: for example, for the request of 'Gadyach' this factor will be more than for the request of 'Moscow Peter Yekaterinburg Samara'.

Long text without links.

This is a muzzle.

Coincidence of thematic spectra of request and document. Request themes-the result of work ((http://wiki.yandex-team.ru/evgenijjkroxalev/subquery Rules of the sorcerer Subquerysearch)) The subject of the document is taken from Yandex-Catalog

The value of the news detector calculated in the Hippo. Always 0 with a detector value less than the threshold.

The quality of incoming links (the classifier of the bream) is broken, cm [405]

The number of incoming links. Remembrance.

The popularity of the request

The language of the document is Russian.

The time of adding a page, more - a more old document; The root is placed from time displayed at the interval [0.1] so that 3+ years gives 1.

If the main page of the owner (most often a second -level domain, for example xxxx.ru), then the factor is 1. For bums, hosting, personal blogs, etc. (for example, Lifejornal, People.ru, etc.) - domains of the third level (such as xxxxx.narod.ru) will also have an equal factor 1.

The time for adding the main page of the owner (host?) Will be remaped like Addtime.

How often they click in this URL for this request - CTR blasting for the correction factor

Simple BM25 in text.

Simple BM25 for links, the weights of the braces are not taken into account.

Simple BM25 in text and links at the same time.

All the words of the request are in the text + links.

Simple BM25 in pairs of words - we take all pairs of words of the request and consider the number of their entry into the text of the document. In the quality of the weight of the pair we use the sum of the scales of words. It does not work if there is a stop-word in the request

The same as txtpair, but for links; Link weights are not taken into account.

BM25 from the number of sentences in the document in which it occurs.

BM25 according to only in the heading.

BM25 according to only with High Rel-bots ('significant', with the allocation (<b> ITP)).

Min (number of words of request/10, 1.f)

1 / quantity_lov_v_

For this Urla, for this request, there is no information about clickness 1 - request or request -URLA in the click database, 0 - query URL in the clicks database

For this request, there is no information about the clickness of 1 - there is no request in the click database, 0 - the request is in the click database.

The number of hops of Url inpans (such as less - closer to the muzzle, the lower the value (0 - the muzzle, 1 - from the muzzle cannot be reached, 0 <can get from the muzzle <1). Normal value for the root of the nosta 0.0039).

the presence of pairs of words in the exact form

the number of sentences in which there are many words in the exact form

the presence of words in the header in the exact form

BM25 in the exact form

Simple BM25 in the exact form.

the presence of pairs of words taking into account synonyms (> = txtpair)

the number of sentences in which there are many words taking into account synonyms

the presence of words in the header, taking into account synonyms

BM25 taking into account synonyms

Simple BM25 taking into account synonyms.

How often they click in the URLs of this Domainid for this request - Ctr Domainid blasting for the correction factor

For this Domainid, for this request there is no information about clickability 1 - request or request -owner is not in the clicks database, 0 - the request for clicks is in the database of clicks

The owner's clickness regardless of the request

It is equal to one if the site has a Ukrainian geoist (i.e. 1 - Ukrainian site)

Page from the blogochosting

Page with Livejournal.com

The quality of the text. It is considered a rather complex formula

Text quality (classifier Alekseeva)

The core of the audience of owners according to Yandex.Mrazusing

The core of the audience of the hosts according to Yandex.Mrazusing

Does the host have a host

The musicality of the request. The results of the sorcerer Anton Konygin.

Document length in sentences

The length of the URL, divided by 5

The commercial request for the dictionary of phrases from Direct: 0 - maximum commercial, 1 - minimal.

The size of the host named after Raskovalov in the documents without taking into account the takes (each double is taken into account in the factor by an independent document)

Document type - HTML

Means the coincidence of the region mentioned in the request and found sites at the level of areas. Binar factor: 1-rush, 0-no. It is based on ((http://wiki.yandex-team.ru/ Yandexposisk/ Classification of Sytraitniki/ Geographic/Sospolzanievpoysk Geoklassification of sites)))))))

Are there any words from Yweb/Pornofilter/Porno.query.

Document from porn kitski

A document from a commercial clay. Not used (depreded)

Fast document

The page title contains commercial vocabulary. Not used (depreded)

page from ru.wikipedia.org

Commercial page (Classifier Savina)

The document does not have all the words of the request (with an accuracy to a synonym)

The percentage of the words of the request in the document (with an accuracy to a synonym)

The document has all the words of the request (with an accuracy to a synonym)

The percentage of the words of the request in the links (with an accuracy to a synonym)

There are all the words of the request in the links (with an accuracy to a synonym)

The value of the commerce detector calculated in the Hippo.

Tr by pairs of words in the reverse order

Lr by pairs of words of the request in the reverse order

TR by pairs of words of the request through one word in texts

Lr by pairs of words of the request through one word in texts

The percentage of all the words of the request in the text (with an accuracy to the form)

The document has all the words of the request (with an accuracy to the form)

The degree of centralization of the points from which the request is set

Whether the request of blog vocabulary contains

Binar non -profit request: Querynoncommerciality> 0.965.

The length of the page text in the words tlen = map (number of words, 1/400), where map (x, y) = x*y / (1 + x*y)

The page is unattainable by the links from the muzzle.

the ratio of the number of clicks on this Urlu to all clicks on request

the ratio of the number of clicks on this Domainid to all clicks on request

[Bug: A copy of factor 45] How often they click in this URL for this request - CTR blasting for a correction factor

What part (on average by the session) from the user Urlov’s user, this URL user, who has been completed to it, is this URL. It is considered to be user sessions.

How often do they click in this URL for this request - CTR blasting for the correction factor, by small regions from Relev_regions.web.txt

How often they click in the URLs of this Domainid for this request - Ctr Domainid to the correction factor, by small regions from Relev_regions.web.txt

The ratio of the number of clicks on this Urlu to all clicks on request, by small regions from Relev_regions.web.txt

The ratio of the number of clicks on this Domainid to all clicks on request, by small regions from Relev_regions.web.txt

Query URL Clicks Combo, in small regions from Relev_regions.web.txt

Query Download Clicks Combo, in small regions from Relev_regions.web.txt

Options for relevant factors taking into account the feet of words

Tr with discount for suggestions

URL satisfies forum_detector regularly

The URL has an ancient date. Ancient news are recognized. Factor 1 if there is a year in Url <= 2007.

The page has a about 'payment SMS'.

Stage of the page

Pornography of the page

Search traffic - transitions from search engines to the site (2nd formula)

Search traffic - transitions from search engines to the site (2nd formula)

The largest total tuning of Urla and request, normalized by the length of Urla

All coincidences are only in the URL, there are no coincidences in the text

Three levels of coincidence of the geography of the user and page

Geographical proximity

Is the request for navigation, on the clicking of the answers

The average for users Active continuous time of the user is (in second) on the host pages after the transition on request from the search engine (the factor depends on the pair (request, Domattr)).

The average for users Active continuous time of the user is (in second) on the host pages after the transition on request from the search engine (the factor depends on the pair (request, Domattr)). In the inside of the Yandex. Bara/elements/browser counter

The average for users is the number of active actions (clicks, clicks) with the continuous finding of the user on the host pages after the transition on request from the search engine (the factor depends on the pair (request, Domattr)). . In the inside of the Yandex. Bara/elements/browser counter

The number of unique visitors from search engines for a specific request

The average for users Active continuous time of the user (in second) on the page after the transition on request from the search engine (the factor depends on the pair (request, URL)).

The average for users Active continuous time of the user (in second) on the page after the transition on request from the search engine (the factor depends on the pair (request, URL)). In the inside of the Yandex. Bara/elements/browser counter

The average for users is the number of active actions (clicks, keystrokes) on the page after the transition on request from the search engine (the factor depends on the pair (request, URL))

Request is not in Russian

The size of the page of the page

The factor is inversely proportional to the size of the page region

The size of the region of the request

The factor is inversely proportional to the size of the regional region

The geographical proximity of the user and the site

The number of unique visitors, remarks exponentially

The share of traffic from search engines

The share of suits to the site is not by links (set with hands or from bookmarks)

average for users Active continuous time for user finding (in sec) on the host pages

The average for users Active continuous time of the user (in second) on the pages of the host. In the inside of the Yandex. Bara/elements/browser counter

The average for users is the number of active actions (clicks, clicks) with the continuous finding of the user (in second) on the pages of the host.

Implementation of the algorithm described in the article ((http://wiki.yandex-team.ru//h.yandex.net/?http%3A%2F%2FreseRosoft.microsoft.com%2fen-US%2FPEOPLIULIUUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUUP032-LIUUUU .pdf http://research.microsoft.com/en-us/people/tyliu/fp032-liu.pdf))

Varla's attendance according to I-Bara

The number of unique visitors to Urla

The average for users time is the user on the page. It is read as the difference between neighboring transitions.

This is the SEA factor = s4_r/ (k_r+10) where S4_R is the number of clicks> 180 sec, k_r - the total number of clicks. It is considered taking into account reformalization.

This is the SEA factor = s4_r/ (k_r+10) where S4_R is the number of clicks> 180 sec, k_r - the total number of clicks. It is considered taking into account reformalization. Localized version

The degree of variety of requests for which this Urla click

IDF for various parts of the document, broken, are not used

IDF for various parts of the document, broken, are not used

IDF for various parts of the document, broken, are not used

IDF for various parts of the document, broken, are not used

BM25 for the user region for localized queries, for the unflapped in Cuba, is a country. The texts of the queries sent for the regions can be viewed in Relev_regions.txt in the sorcerer

The same for lingonic relevance

The share of incoming corrupt links. The algorithm for recognition of commercial links is implemented. The factor will be remarked to [0.1] if the share of such links is 50%, otherwise 0. ((http://wiki.yandex-team.ru/svetlanashorina/topseolinks selection of wound sites))))))

Previous factor multiplied by Pornoquery

ComMlinksseohosts factor multiplied by Noncommercialquery

The request mentions the product category. Not used (depreded)

The request mentions a vendor. Not used (depreded)

Geographical distribution of the request

The request is set mainly at night

The request is set mainly in the morning

The request is given mainly in the afternoon

The request is set mainly in the evening

The severity of the querial tasks at different times of the day

Coincidence of thematic spectra of request and document. Request themes - the result of work ((http://wiki.yandex-team.ru/evgenijjkroxalev/subquery Rules of the sorcerer Subquerysearch)) The subject of the document is determined by the automatic classifier

Entropy - distribution of clicks

Entropy - distribution of shows

Entropy - Distribution of clique/shows.

Classifier of Porn Causions, another dictionary than Pornoquery

The geographical proximity of the country of the site and the country of request

Coating domain three -bouqu and request. (Chelyabinsk lottery - Chelloto. We translate a request to translite, find the three -book that are covered (Che, Hel, Lot, Olo), we look at what share of all three -bouquets are covered)

The same as the previous factor, but about the entire Url except the domain

The request is local-specific. The request is often reformulated with the obvious task of the region. ((https://ml.yandex-team.ru/archive/thread1433892/#Message1433892 more))

Analogs of the corresponding text factors for links. BM25 from the number of links in which a coincidence occurred.

Simple BM25 in the exact form in link texts

The presence of pairs in the links of the words, taking into account synonyms

The number of links passed the threshold

Simple BM25 by links taking into account synonyms

Request about the video

The owner's clickness regardless of the request, separately in the regions

Entropy is the distribution of clicks. Regionalized

Entropy is the distribution of shows. Regionalized

Entropy - distribution of clique/shows. Regionalized

equals 2 * NastyContent

equals 2 * NastyContent

Domna in Zone .com

Domain in the .ua zone

Domain is not in the .ru zone

The poetry of the document

The maximum poetry of the quatrain

Document language - English

The language of the document is Cyrillic

Requestful factors - the result of work ((http://wiki.yandex-team.ru/jandekspoisk/kachestvopoiska/obshayaformula/tekushhiekomponenty/georegionality classifier of geolocalization of the request)))))))))))))

R- Georelevan - regional results in the issuance could be useful, but nothing more

V- geovital - regional issuance is of fundamental importance

There are no numbers in Urla

Show how much the text is unnatural from the point of view of the Russian language. Assessment of how much the text of the document can be considered as a generated synonymizer or automatic. ((http://wiki.yandex-team.ru/jandekspoisk/kachestvopoiska/obshayAfermula/tekushhiekomponenty/antispam?v=1il#h58953-2 more))

Show how much the text is unnatural from the point of view of the Russian language. Assessment of how much the text of the document can be considered as a generated synonymizer or automatic. ((http://wiki.yandex-team.ru/jandekspoisk/kachestvopoiska/obshayAfermula/tekushhiekomponenty/antispam?v=1il#h58953-2 more))

Show how much the text is unnatural from the point of view of the Russian language. Assessment of how much the text of the document can be considered as a generated synonymizer or automatic. ((http://wiki.yandex-team.ru/jandekspoisk/kachestvopoiska/obshayAfermula/tekushhiekomponenty/antispam?v=1il#h58953-2 more))

ND/K normalized time to click

nd/i

CONTRY / K

w/k

o/i

Selected formula

r_s4b/(r_k + 10)

Does the request have a complete syntactic analysis

The date of the document that is registered on the page is remarkable

The document has textual relevance

An indicator of the unnaturalness of the text from the point of view of the Russian language. The number of bad pairs of words in the text, transferred to the segment [0.1] according to the Z/(Z+10) formula

The proportion of bad steam among all found in the table: Z/(X+1), where Z is the number of bad couples in the text, and X is (http://wiki.yandex-team.ru/evgenijgrechnikov/testsynonimizers of 2000-navigable )) steam

The number of Latin letters in the text (not counting the markings), driven into [0.1] formula n/(n+100)

Previous factors - fixed

Previous factors - fixed

Previous factors - fixed

Previous factors - fixed

factor cunningly combined from FRC and Pseudo-CTR

factor cunningly combined from FRC and Pseudo-CTR

The number of words in the text (the word is what the lemmeter selected) is displayed in [0.1] according to the formula x/(x+a)

The number of words of the Russian language in the title

The average length of the word

The percentage of the number of words inside the tag <a> .. </a> from the number of all words

The percentage of the number of words outside the tags (outside the brackets <>) from the number of all words

The percentage of the number of words, which are 200 the most frequent words of the language, from the number of all words of the text

The number used in the text 500 of the most popular words of the language, divided by 500

Logarithm of average geometric probabilities of trigrams in the text. (the probability of a trigram - the number of its meetings in the text, divided by the number of all trigrams) is displayed in [0.1] according to the formula -x (x+a)

Logarithm of the average geometric conditional probabilities of trigrams. The conditional probability of a trigram is its probability, divided by the probability of a bigram from the first two words

The analogue of the QueryDownerClickSpCTR factor differs from it in that the requests are normalized by doppelgage (details of such normalization -((http://staff.yandex-team.ru/finder Andrei Plakhov)), code/yandex/doppelganges)

The analogue of the QueryDownerClickspCTR factor differs from it in that the requests are normalized according to doppelgage (details of such normalization -((http://staff.yandex-team.ru/finder Andrei Plakhov)), code/yandex/Doppelganges). Localized to Relev_regions.web.txt

The analogue of the QueryurlClickSpCTR factor differs from it in that the requests are normalized by doppelgagers (details of such normalization - ((http://staff.yandex-team.ru/finder Andrei Plakhov)), code - Yandex/Doppelganges)

The analogue of the QueryurlClickSpCTR factor differs from it in that the requests are normalized by doppelgage (details of such normalization - ((http://staff.yandex-team.ru/finder Andrei Plakhov)), code - Yandex/Doppelganges). Localized to Relev_regions.web.txt

BM25 on URL'U

The page has a big picture

The difference between the current date and the date of the document defined by the dates, 1 - the date of the document is equal to the current, 0 - the document of 10 years or more, if the date is not defined, equal to 0. Attention! ((1 - dateraage)*60)^2 = age of the page In days.

The number of incoming links without Russian letters. Remembrance.

The maximum number of forms in all words of the request is max in all words of the request request_form_dl_lov/64

The sum of the number of forms balanced by the scales of words - the amount in all words of the request of the number_form_dly_lov/64*weight_lov; REMAP species x/(1 + x).

The unwarmed amount of the number of forms is the amount in all words of the request of the number_form_dl_lov/64/number_lov_

The maximum number of forms in all words of the request

Summer of the number of forms balanced by scales

Undested amount of the number of forms

Analogues of the factors of the same name, the weight of the word = 1

Analogues of the factors of the same name, the weight of the word = 1

Analogues of the factors of the same name, the weight of the word = 1

Analogues of the factors of the same name, the weight of the word = 1

The share of different parts of speech in the text. The share of numerals (among all words that managed to recognize part of the speech)

The share of particles

The share of pronoun adjectives

The proportion of pronoun nouns

The share of verbs

The share of words that can be both masculine nouns and nouns of the feminine, but not of the middle kind, among all nouns (examples: 'hummingbirds' are an example of an indefinite kind that can be determined in two ways, 'Alexander' is homonymy).

Quality of incoming links (hauser classifier) ​​corrected

Considered LinkQuality for this page or not (did not think, if there are few links) corrected

Quality classifier of incoming links 2 corrected

The request is the name of the organization (example: Gazprom, Gazprom) ((http://wiki.yandex-team.ru/arsengadzhikurbanov/warees Description))

The size of the largest text segment (from the factor [18] puretext)

The number of different internal links to the page

The city is defined for the site

Requestful factors - the result of the work ((http://wiki.yandex-team.ru/poiskovajaplatforma/lingvistika/zaprosnyjefefactory/localizovannyjezaprosya classifier of the request of the request)) - a new version of factors [328] - [328] - [328]: u - u - u - u - u - u - uceleless sites the request is meaningless;

Запросные факторы - результат работы ((http://wiki.yandex-team.ru/PoiskovajaPlatforma/Lingvistika/ZaprosnyjeFactory/LocalizovannyjeZaprosy классификатора геолокализованности запроса)) - новая версия факторов [328]-[330]: R - георелевантные - региональные результаты в issuing could be useful, but nothing more;

Requestful factors - the result of work ((http://wiki.yandex-team.ru/poiskovajaplatforma/lingvistika/zaprosnyjefefactory/localizovannyjezaprosya classifier of the request of the request)) - a new version of factors [328]: Vegetable fundamental importance.

Ukrainian Page Rank

= 1 - v. Download formula. Class requests: download/watch online/play/photo/listen

The result of the classifier of the request - in the request there are words from the corresponding dictionary. brand

Medication Dictionary

question

Specific request for Moscow

organization

porn

trips

The frequency of links to the site

The number of almost-periodic links

The number of Owner shows on request, normalization x/(100 + x).

The number of URL shows on request, normalization x/(100 + x).

The presence of a LiveInternet meter

The popularity of Owner is in requests

The value of the pirate detector calculated in the hippo.

It is considered as (10-x) where X is the return of the document in days (continuously) regarding the validity time of the document in Samovar

The host of the document is recognized in the request

URL consists only of the host, which is recognized in the request

URL is the plot of Yandex News

URL feature computed from rapid clicks spy_log counters with decay of 1 day

URL feature computed from rapid clicks spy_log counters with decay of 1 day

URL feature computed from rapid clicks spy_log counters with decay of 0.5 days

URL feature computed from rapid clicks spy_log counters with decay of 0.5 day

Simple BM25 in pairs of words - we take all pairs of words of the request and consider the number of their entry into the text of the document. Weight = 1. It does not work if there is a stop-word in the request

Logarithm of the number of shingles on which this document is not unique

Logarithm of the number of shingles on which this owner of the document is recognized as the author

The average weight of non-ugly shingles of this document

It reads in the same way as the Hostrank factor, but not on all the Owner graph, but on its subrack, consisting of Owner's in this region. Belonging to the region is determined by TLD, or by the presence of pages with this Owner in the index, about which the GEO or Geoa classifier says that they are from this region. Mapped in the same way as the Hostrank factor, from 0 to 1 with 256 gradations

A document from the language section of Wikipedia corresponding to the user region

The popularity of the request within the country

The degree of centralization of the points from which the request is set (inside the country)

Geographical distribution of the request within the country

The hour at which this request is given the most

The degree of severity of the querial tasks at different times of the day (inside the country)

The country of the document (domain) and the country of the user coincide ((http://wiki.yandex-team.ru/jandekspoisk/kachestvopoiska/obshayaFormula/tekushhiekomponijafaktorov#national

On the Porn Advertising page

URL feature computed from rapid clicks spy_log counters with decay of 3 days

Country classifier of localization - how much the request implies the context of the country

The number of slashes in Url

BM25 with different parameters for different fields, including an incoming anchortekst. The weight of the text of the links included on the page is normalized depending on Delta Page Rank links

The presence of a built -in video player on the page

Video for downloading

URL feature computed from rapid clicks spy_log counters with decay of 3 days

URL feature computed from rapid clicks spy_log counters with decay of 14 days

The factor is calculated from the text of Url using the classifier of sequences Quality/Seq/GSK

Model with the training of each trigram on '+' and '-' Urlah. It does not depend on the request.

URL feature computed from rapid clicks spy_log counters with decay of 14 days

Age of rapid clicks spy_log update, in seconds

Freshness of rapid clicks spy_log update

The measure of 'commercial' request. It is a comprehensively calculated Matrixnet factor formula for the procurement vocabulary in direct + for user queries + add. Intensive dictionaries. Requests with intensity to buy a factor seeks to -> 1 commodity requests -> 0.6 with intensity cannot buy, reviews, etc. -> 0 ((http://wiki.yandex-team.ru/Faktorydljanovogokatorazaprosov Factors of the Classifier))) (HTTP : //wiki.yandex-team.ru/jandekspoisk/antispam/antiseo/klassifikATORCHESSKIXZAPROSOV STUNITURE OF HIM))

Unigramal language model. Language is modeling according to the document, smoothed out by the general linguistic model. When building a model, the document uses information on which field of the document met the word request (Title, Head or Plain Text)

The coincidence of geography, determined from the Url of the document and the city of the request (IP or LR)

The coincidence of geography, determined from the Url of the Document and the Request region (IP or LR)

The coincidence of geography, determined from the Url of the document and the country of request (IP or LR). Actual for Russia and Ukraine.

The coincidence of geography, determined from Ural Documents and the City in the request (GEOCITY rule)

The value of the chopped commerce detector, calculated in the hippo.

The share of clicks for navigation requests

The result has a geography of the user at the city level ([415] == 1 && [215] == 0)

Request geovitality for results from the user region

Request geovitality for the results is not from the user region

The share of the Urlov that respond without errors

Coincidence of the thematic spectrum (according to DMOZ) request and document. The theme of the request is determined ((http://wiki.yandex-team.ru/jandekspoisk/zarubezhnyjjinternet/dmozqueryClassifier1 The rule of the sorcerer Dmoztheme))

Coincidence of the thematic spectrum (according to DMOZ) request and document. The theme of the request is determined by the best result ((http://wiki.yandex-team.ru/jandekspoisk/zarubezhnyjjinternet/dmozqueryClassifier1 Rules for the sorcerer DmozTheme)) The subject of the document is determined by the automatic classifier

The size of the page of the page

Freshness of rapid clicks spy_log update, calculated at the request time

It launches on the basic search under the name ISGEO the maximum weight of the meters of the gelator in the request. A geo-object is understood as an object of the category GEO, Geo1, Geoaddr, Geoaddr1, Landmark, Landmark1 (see ((http://wiki.yandex-team.ru/alekseysokirko/queryobjects kaovsky allocation))))))))))))))))))))))))))))))). wiki.yandex-team.ru/arsengadzhikurbanov/wares Read more))

It launches for the basic search under the name ISMUSIC the maximum weight of the Music or Music1 category of the category of the Category in the request. (See ((http://wiki.yandex-team.ru/alekseysokirko/queryobjects SOM-OV)))). ((http://wiki.yandex-team.ru/arsengadzhikurbanov/warees more)))))))))))))))))))))

Modification of the BCLM2 factor, lightweight for use in tulle. The main difference is that BCLMLite does not use absolute displacements of words relative to the beginning of the document. Instead, the factor works with the usual positions of the type <number of the_prising, position_v_production>. At the same time, the proximity between the words is taken into account only inside the sentence. (Http://wiki.yandex-team.ru/jandekspoisk/kachestvopoiska/obshayaFormula/tekushichiekomponenty/bclmlite bclmlite)))))))))))))

When responding to a request, the results are important in close proximity ([pharmacies], [children's clinic])

When answering a request, the results within the city are important (the bulk of localized queries)

When responding to a request, the results from the region of the user ([airport], [dairy]) are important

The number of incoming muzzle links

Page - '404' (share of tokens '404' in relation to the total number of tokens on the page)

URL feature computed at the request time from rapid clicks spy_log counters with decay of 1 day

ND/K normalized time to click

URL feature computed at the request time from rapid clicks spy_log counters with decay of 0.5 days

CONTRY / K

w/k

o/i

Selected formula

r_s4b/(r_k + 10)

Number of letters in the AUX segment

The number of spaces in the AUX segment

The number of commas in the Content segment

Page is a store. ((http://wiki.yandex-team.ru/jandekspoisk/kachestvopoiska/obshayAformula/tekushhiekomponenty/opisanijafaktorov#SSHOP Description)). Not used (depreded)

Urlngramsmodel ranking factor in ERF

The language of the document corresponds to the country's request

The degree of variety of requests for which this Urla click is read by regions

nd/i

The language of the document is one of the permissible for Turkey (Turkish, English, German, French, Arabic, Azerbaijani) or the document has zero length. In the search stage is calculated only for Isrealgeolocal requests.

Document links dispersion

The number of clicks on the owner and the number of clicks on request more than 5

BM25FDPR with standardization on the average length of the document, depending on the language of the document. ((http://wiki.yandex-team.ru/bm25frework test results.))

The popularity of the language of the document. Number from 0 to 1. (http://wiki.yandex-team.ru/jandekspoisk/kachestvopoiska/obshayaformula/tekushhiekomponenty/languaguaguagepopalarity)))))))

The amount of factors QueryDownerClicksFRC and BM25FDPRFIXED with scales 0.358449 and 0.184922, respectively. '565' in the name of the factor does not need to be perceived literally, it is Legashi or a typo.

URL feature computed at the request time from rapid clicks spy_log counters with decay of 3 days

URL feature computed at the request time from rapid clicks spy_log counters with decay of 14 days

Dispersion of languages ​​in XMAP

There is a typo in the request

The ratio of the number of incoming links, the text of which is the URL, is one of the incoming links

A pool of logs is marked with BERT trained on Sinsig. DSSM model is trained on this pool using BaseregionChain

The number of 'Nebukv 'in Url

The length of the URL with an accuracy to the symbol. Disconnected in production.

Habi page

BM25 page title by its text

BM25 page title by texts of links to it

The number of COO-Thrilling links between hosts

Static URL factor for search sessions for 1600 days calculated on mobile sessions. Average dwelltime, and DwellTime is cut from the session if more than 180 seconds

Static URL factor for search sessions for 1600 days calculated on mobile sessions. The probability that the URL click will be more than 120 seconds

Static URL factor for search sessions for 1600 days calculated on mobile sessions. The probability that URL is not clicks if they click at least one URL below.

Static URL factor for search sessions for 1600 days calculated on mobile sessions. Average dwelltime, and DwellTime is cut from the session if more than 3600 seconds. Localization to the level of countries.

Static URL factor for search sessions for 1600 days calculated on mobile sessions. Average dwelltime, and DwellTime is cut from the session if more than 180 seconds. Localization to the level of countries.

The value of the health detector calculated in the Hippo.

The share of unique trigrams in the trigrams of links

Share of unique trigrams of links in trigrams header

The greasy of the page

Similar to Yabarurlvisits

Regional and navigation request - in the user region there are one or more navigation results on it

The number of sessions in which Url was the last, classified as the sessions in which Url appeared

The sum of the maximum values ​​of Sourcerank's for each incoming link, taking into account the uniqueness of the owner.

The weight of the words of the request that is in the text in the exact form

The weight of the words of the request that is in the text with an accuracy to lemma

The weight of the words of the request that is in the text

It launches on the basic search under the name ISHUM the maximum weight of the enclosed object of the Hum or Hum1 category in the request. (See ((http://wiki.yandex-team.ru/alekseysokirko/queryobjects SOM-OV)))). ((http://wiki.yandex-team.ru/arsengadzhikurbanov/Wares#ishum more)))))

It launches on the basic search under the name ISTEXT the maximum weight of the TEXT or Text1 category of the category of the category met in the request. (See ((http://wiki.yandex-team.ru/alekseysokirko/queryobjects SOM-OV)))). ((http://wiki.yandex-team.ru/arsengadzhikurbanov/Wares#istext more)))

It launches on the basic search under the name Ispicture the maximum weight of the Picture or Picture1 category of the category of the category of the category in the request. (See ((http://wiki.yandex-team.ru/alekseysokirko/queryobjects SOM-OV)))). ((http://wiki.yandex-team.ru/arsengadzhikurbanov/Wares#ispicture))))))))))))))))))

Returns the maximum degree of household objects in the request under the name Wmaxone. (See ((http://wiki.yandex-team.ru/alekseysokirko/queryobjects SOM-OV)))). ((http://wiki.yandex-team.ru/arsengadzhikurbanov/Wares#maxone more)))))))

Returns the maximum degree of household objects in the request under the name Wminone. (See ((http://wiki.yandex-team.ru/alekseysokirko/queryobjects SOM-OV)))). ((http://wiki.yandex-team.ru/arsengadzhikurbanov/Wares#minone more)))))

Allows you to evaluate whether the document is 'alive' is from the point of view of links to it coming.

Similar to Yabarurlvisitors

Similar to Yabarurlavgtime

The share of clicks on this Urlu among all clicks on similar requests

Corrected CTR of this Urla for all similar requests

URL feature computed at the request time from rapid clicks search counters with decay of 1 day

URL feature computed at the request time from rapid clicks search counters with decay of 30 days

The most likely theme of the request determined ((http://wiki.yandex-team.ru/jandekspoisk/zarubezhnyjjinternet/dmozqueryClassifier1 The rule of the sorcerer DmozTheme)), only the most popular topics are taken into account (but there are more than in the DMOZQUREMES factor). The factor contains the likelihood of a correspondence of the request of the theme, but for each topic, its own interval is taken on the segment [0..1]

The theme of the request determined ((http://wiki.yandex-team.ru/jandekspoisk/zarubezhnyjjinternet/dmozqueryClassifier1 The rule of the sorcerer Dmoztheme)), only a few of the most popular topics are taken into account.

0 or 1, depending on the presence in the request of the clearly expressed intent Need_photo from the variety

0 or 1, depending on the presence in the request of the clearly expressed intent Need_map from the variety

The factor is an analogue of LongQuery (the sum of the IDF words of the request), but with the 'correct' accounting of synonyms. Specifically, a minimum of IDF (i.e. the most frequent) of synonyms and words is selected.

Url contains a token that coincides with the short name of the user country. The factor is considered only on the EU stream.

Expected number of found on request

The share of unique trigrams of a footer fragment in trigrams of links

The share of unique trigrams of links among a fragment of trigrams of a footer

Double logarithm of the probability of a request for a language model of the Erratum typo service

'Fixed' clicks counted using REQUESTAGGRETELIB

'Fixed' clicks calculated using Requestaggregatelib. Regional version

Regional attendance of Urla according to the I-Bara

The average time of the user stay on the host with an external (from another non-search site) entry from a specific URL

The average 'depth' (the number of transitions within the framework of the host) of the user stay on the host with an external (from another non-playing site) entry from a specific URL

The average length of the logical session in which there was a request

Assessment of the probability of leaps from the document

User outflow coefficient from the search after a visit to the site

This is Index. (HTML/PHP/ASPX?/...), without CGI parameters. It is considered to be for all takes.

This is Index. (HTML/PHP/ASPX?/...), possibly with CGI parameters. It is considered to be for all takes.

Whether the host is the owner, conditionally host == Owner (Host).

The minimum length of Pathandquery for all half -shoes.

Classifier for estimates of commercial sites

The document has a direct link to the file

The document has a link to filehosting

0 or 1 - whether the request is matured by the tickt

0 or 1 - whether the request is matured by the tickt

0 or 1 - whether the request is matured by the tickt

The prediction of the share of “good” (at least two different cities and frequency> = 10) references to the request with geography in Turkey

The result of the work of the lexical classifier of requests predicting the likelihood of click on the theme of 3561

The result of the work of the lexical classifier of requests predicting the likelihood of click on the page 3973 page

Rank 'navigation'

Regional attendance from search engines for a specific request

Clicks on the urlahs shown in the issuance for requests, by which they went to look for other search engines

Urlov shows in the issuance for requests, by which they went to look for other search engines

Classifier of the commerciality of the site

The document contains user review/comment

The share of clicks on this Urlu among all clicks according to similar requests, the country version, see ((http://wiki.yandex-team.ru/development/poisk/Arcadia/indexregex Indexregex)))))))))

Corrected CTR of this Urla for all similar requests, country version, see (http://wiki.yandex-team.ru/development/poisk/Arcadia/indexregex Indexregex))))))

The average number of found on request

The angle in the Depth Nodes space, counted only by words (min for all)

Classifier that approximate the quality of commercial sites based on user behavior data

The time of creating a document with an accuracy of 1.0 is the current month, 0- 10 years ago and older. Temporarily disconnected

The time for updating the document with an accuracy of 1.0 is the current month, 0- 10 years ago and older. Temporarily disconnected

The function of the credibility of the distribution of years in the document. Temporarily disconnected

The arithmetic mean position of dates in the document. Temporarily disconnected

The average position of Urla for a normalized request

The average position of Domattr for a normalized request

The average position of Urla for all requests

The average position of the host for all requests

Number of requests for URL

Number of requests for host

Implementation of the algorithm described in the article ((http://wiki.yandex-team.ru//h.yandex.net/?http%3A%2F%2FreseRosoft.microsoft.com%2Fen-US%2FPEOPLIULIUUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUULYUUP032-LIUUUU .pdf http://research.microsoft.com/en-us/people/tyliu/fp032-liu.pdf)) by large regions (tube)

The share of the words of the document from the segments with Score> 2.

The rank of site quality used for boosts of the Moscow commercial formula

The factor is used in Selectionrank. TG_UNUSED: should not be included in the formulas to avoid feedback

URL feature computed at the request time from rapid clicks search counters with decay of 3 days

The weight of the document on a monosyllabic dictionary of commercial vocabulary

Average query commerciality

The number of requests in the group of frequency requests similar to a given

FRC groups of frequency requests similar to a given, with averaging through the sum of clicks and shows

FRC groups of frequency requests similar to a given, with averaging through the sum of clicks and shows, according to regional statistics

URL feature computed from rapid clicks search frozen counters with decay of 1 day

The share of clicks on this Urlu among all clicks on similar requests, counted according to Popular Search Engine

The length of the Depth Nodes petal counted for hosts

Dispersion of the angle in the space of Nodes Time, calculated for hosts

0.9-quarter of the length of the petal in the space of Nodes Time, calculated for hosts

Content ugliness factor.

CTR according to click data, the request is normalized according to Sinsets

Regional CTR according to click data, the request is normalized according to Sinsets

Static trigrams intercection of url and queries by which users visited the url.

The result of the work of Adult Rules for the Sorcerer.

Weighed BM15 for a request for an index document - a list of requests for which they switched to it.

The likelihood of a racing from a host after click (on the logs of the bar).

The number of chains on request / (the number of chains in which URL + the number of chains on request participated).

The number of chains in which Url was the last normalized for the total number of chains in which this URL was.

Number of transitions to URL from Wikipedia

URL feature computed from rapid clicks search frozen counters with decay of 30 days

The page indicator is like a hub (how many pages are the bar users pass from it).

TEXTBM25 is considered in the title by the text of the name of the user region - similar to the factor 268.

DSSM Prediction of the probability of URL + Title that there is no product on the page.

FRC Popular Search System for Browser Logs

Weighted mean of log(query_clicks)/log(query_shows) for given host. Weights are proportional to log(query_shows) + 0.2.

The number of transitions on the request for URL, found in the Hopes chain, normalized to the general garlic of the transitions on request.

The probability of Urla to be the last upon request in the chain of Hopes.

DSSM Prediction of the probability of URL + Title, which is on the page one product.

DSSM Prediction of the probability of URL + Title, that there are a lot of goods on the page.

URL feature computed from rapid clicks search frozen counters with decay of 3 days

For Urla, a geo-approval of the city level is determined according to the rules of the BUKI-1125

For Urla, a geo-approval of the country's level is determined according to the BUKI-1125 rules

Factor Gorelevregions of the 1th Attichut and Geoa

Factor GorelevregionRegionRegion Natthew GEOA

Factor Geogeetryproxim ▪ Attributu GEOA

Factor Gorelevaliencity n Att. Att. Attibtu Geoa

Factor Geovqueryinusercidence n Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Att. Attfruut and Geoa

Geovquery Geovqueryinieniencity n Att. Att. Attib

PageRegionsize Factor by GEO attribute

PageRegioncoverage Factor GEO attribute

PageRegioncoverage Factor on Adresa attribute

GeorelevregionCity Factor on Adresa attribute

What part (on average in the session) from the clinked in this query Urlov is this URL. It is considered to be user sessions.

Aries is a store

Aries is a service

FRC by transitions from requests that were set by the user several times

The average weight of the shows on the first page; Click weighs 1, non -click - according to the SBM_GAMMAS table

The average weight of the shows on the first page; Click weighs 1, non -click - according to the SBM_GAMMAS table. Regional version

Half -Summaria assessment of the position of Url with a median position for all similar queries on bisters

Host feature computed at the request time from rapid clicks spy_log counters with decay of 3 days

Host feature computed at the request time from rapid clicks spy_log counters with decay of 3 days

Host feature computed at the request time from rapid clicks spy_log counters with decay of 14 days

Host feature computed at the request time from rapid clicks spy_log counters with decay of 14 days

Host feature computed at the request time from rapid clicks spy_log counters with decay of 3 days

Host feature computed at the request time from rapid clicks spy_log counters with decay of 14 days

Host feature computed from rapid clicks spy_log counters with decay of 3 days

Host feature computed from rapid clicks spy_log counters with decay of 3 days

Host feature computed from rapid clicks spy_log counters with decay of 14 days

Host feature computed from rapid clicks spy_log counters with decay of 14 days

Host feature computed from rapid clicks spy_log counters with decay of 3 days

Host feature computed from rapid clicks spy_log counters with decay of 14 days

Aries is a partner

The document is Shopinshop

The value of the conversion of the request calculated in the Hippo.

At least one offer from a sporled scheme has an accessibility status.

There is not a single offer in the porous scheme.

URL is a product on the market.

URL is a product on the market and has Offerid.

URL is Shopinshopcpa.

At least one offer from a sporled scheme has an inaccessibility status.

DSSM Prediction of the probability of URL + Title that the document is an overlap.

In the offer from the new Parser, the PartnerOfferContent Available field is exhibited.

In an offer from the new Parser, the PartnerOfferContent Available Field == True.

Normalized corrected clicks count by query with user's city(gc=) mentioned

Normalized corrected clicks maximum ratio by query with user's city(gc=) mentioned

Normalized corrected clicks maximum ratio by query with not user's city(gc=) mentioned

The value of PURCHASETOTALPredict, calculated in the Hippo.

The value of serpsummarysurpluspredict, calculated in the hippo.

User return on URL

Requestwith120d3ClickpartPredict value, calculated in the Hippo.

The value of the requester detector of the parasites calculated in the hippo.

Middle Logarithm by the user's location on the host with localization in the country; It is considered according to Yabar logs

The attitude of Dwell Time on a host in this region to Dwell Time on a host in all regions

The attitude of Dwell Time on the page in this region to Dwell Time on a page in all regions

The more users add to bookmarks a url, the more factor value it has

Predict SOS.DSSM models by URL + Title.

Med.DSSM Predictions URL + Title models.

FIN_LAW.DSSM Predictions URL + TITLE.

On danny url is a link from inFobox-ov to Wikipedia.

Predict Cruelty.dssm URL + TITLE models.

The value of Halfecompredict, calculated in the Hippo.

A factor similar to RegexmaxclickPercentreg, but calculated by Preffix-Suffix Generalization.

A factor similar to REGEXMAXCLICKPERCENTYABARREG, but calculated according to PREFFIX-SUFFIX Generalization.

Request and documentary navigation model.

The average sung of inclination in the plane of the peak

Request-murl factor. Value is the result of the collaborative filtration of data for the QueryurlCorrectedCTR factor

The factor on the text of the request and heading (Title) of the document, assessment of the compliance of numerical ranges in words-markers

A sign that the document was received by machine translation

Prediction of Med_with_Trash.DSSM (Medic. Document model with Tresh Valley in Lern) Models for URL + Title.

Prediction FIN_LAW_WITH_TRASH.DSSM (Fin-Jur. Document model with a tresh valve in Lern) Models for URL + Title.

The share of Urla in the total number of Urlov closed for the session on request (Synnorm).

The average share of clicks for this UrLU for this request among all clicks for this request (Synnorm) during the day.

The average share of clicks for this UrLU for this request among all clicks for this request (QNORM) during the day.

QI version of factor 861. MaxValue over the set of popular similar queries.

QI version of factor 798. MaxValue over the set of popular similar queries.

Request-murl factor. Value is the result of the collaborative filtration of data for the SampleperiodDayFRC factor

QI version of factor 879.

A document is a short video (Tiktok, Reels, Shorts).

Document-telegram channel in web format.

Document - post in a telegram.

Factor about the presence of a symbol '?' In Url. It is zero if the Url has CGI parameters (more precisely: all duplicate have a symbol '?' In Url).

Prediction of a share of requests with geography on a bag of words built for request

The request is a Urle with an accuracy of the points and testing characters - the ISURL sorcerer's rule is used

The likelihood that the Yweb/Robot/urlgeo_ml region is correct is correct, provided that the city is predicted

Normalized corrected clicks maximum ratio by query with user's city(gc=) mentioned equal by region

Normalized corrected clicks maximum ratio by query with user's city(gc=) mentioned equal to user's region

The value of the classifier of localization for request

Relev_locale == ru

Relev_locale == ua

relev_locale == by

relev_locale == kz

Porn query classification result from Wizard (iad_vw flag, based on Vowpal Wabbit)

URL coating with trigrams from the request. Analogue of Urldomainfraction, Urlpathandparamsfraction factors.

The share of visits for which the time spent during the day on the host is more than 90 seconds

The share of visits for which the time spent during the day on the host is more than 160 seconds

Rank of hacked sites

Rank AGS4

Maximum QSRANK on the owner

Average QSRANK on the main domain

The share of users who returned within a month

The number of users returning within a month

Rank Dorweev

Share of the capital letters in Title

The share of incoming traffic from search engines among all incoming traffic

The share of direct visits among all incoming traffic

Middle QSRank in a sliding window

Minimum QSRANK

Average HOPS

The binary factor about the mobile adaptability of the document. It is taken from ERF

In those cases when fi_national_domain is 0, and Herf.NationalDomainid is filled 1

A broken built -in video on the page.

Random float in [0,1] by user request and document

the ratio of the total area of ​​all Flash blocks to the screen area

Copy of Old Version No.294 Factor. Added for Use on L3 Stage Only. Coating domain three -bouqu and request. (Chelyabinsk lottery - Chelloto. We translate a request to translite, find the three -book that are covered (Che, Hel, Lot, Olo), we look at what share of all three -bouquets are covered)

Fast version of FI_URL_DOMAIN_FRACTION

Document - this is a selection of Tiktok /Tag

Document - this is a selection of Tictock /Discovery

Document - this is a selection of Tiktok /Music

Request-document model Sinsiga.

Static URL factor in search sessions in 1600 days. Ordinary CTR.

Static URL factor in search sessions in 1600 days. Average dwelltime, and DwellTime is cut from the session if more than 3600 seconds

Static URL factor in search sessions in 1600 days. Average dwelltime, and DwellTime is cut from the session if more than 180 seconds

Static URL factor in search sessions in 1600 days. The probability that the URL click will be more than 120 seconds

Static URL factor in search sessions in 1600 days. Logarithm of the number of shows.

Static URL factor in search sessions in 1600 days. The probability that URL is clicking if they do not click on at least one URL higher.

Static URL factor in search sessions in 1600 days. The probability that URL is not clicks if they click at least one URL below.

Static URL factor in search sessions in 1600 days. Ordinary CTR. Localization to the level of countries.

Static URL factor in search sessions in 1600 days. Average dwelltime, and DwellTime is cut from the session if more than 3600 seconds. Localization to the level of countries.

Static URL factor in search sessions in 1600 days. The probability that the URL click will be more than 120 seconds. Localization to the level of countries.

Static URL factor in search sessions in 1600 days. The average position of the URL for all requests. Localization to the level of countries.

Static URL factor in search sessions in 1600 days. Logarithm of the number of shows. Localization to the level of countries.

DSSM model trained on clicks. Takes bigrams into account. Embeddings for documents are computed offline.

Rank of the quality of texts on the host. The higher, the greater the likelihood that the host is full of articles - a rewriting, a bad copy of the content ordered on the exchanges of content. Burning stronger as the before the aggregation.

The probability that the VK.com host is popular for this request in accordance with the corresponding DSSM model.

The likelihood that the Onliner.by host is popular for this request according to the corresponding DSSM model.

The probability that the Rambler.ru host is popular for this request in accordance with the corresponding DSSM model.

The likelihood that the ExpertCen.ru host is popular for this request in accordance with the corresponding DSSM model.

The probability that the Sunhome.ru host is popular for this request according to the corresponding DSSM model.

Static URL factor in browser logs for the maximum period. The percentage of traffic from social networks in all traffic from other hosts and search.

Static URL factor in browser logs for the maximum period. The average number of direct descendants from the host on which they spent more than 90 seconds. The descendant is straight, only if there is a link from our page to the descendant and crossed it.

Static URL factor in browser logs for the maximum period. The average maximum depth of wood with the root in the current URL is when the URL is visited from other hosts.

Static URL factor in browser logs for the maximum period. The number of times when the feather was transferred to the page to the total number of pages to which they switched from a sickle. The closer to 1, the more often the page was opened the only one in the session.

Static URL factor in browser logs for the maximum period. The average length of search sessions, when they switched to the page from a sickle

Static URL factor in browser logs for the maximum period. The formula for calculating the factor we look at Wiki.

Static URL factor in browser logs for the maximum period. The formula for calculating the factor we look at Wiki.

Static URL factor in browser logs for the maximum period. The probability that the user will spend on the page> 120 seconds.

Static URL factor in browser logs for the maximum period. The number of leaves in URLA support. In this case, the leaves are a page from which there were no transitions.

Static URL factor in browser logs for the maximum period. The average time spent on the page and in all descendants of the page (URLS to which they switched) from the host. Cut off if the total DT is more than 10 minutes

Static URL factor in browser logs for the maximum period. The minimum Unix Time when the page appeared in the logs for the first time.

Static URL factor in browser logs for the maximum period. The difference between the middle and minimum Unix Time when the page appeared in the logs.

Static URL factor in browser logs for the maximum period. Current breadth where the page was viewed from.

Static URL factor in browser logs for the maximum period. Current longitude where the page was viewed from.

Static URL factor in browser logs for the maximum period. The likelihood of leaps from the page

Static URL factor in browser logs for the maximum period. The likelihood of image jumps from the page

Static URL factor in browser logs for the maximum period. The probability of leap torrent file from the page

Static URL factor in browser logs for the maximum period. The formula for calculating the factor we look at Wiki. Localization to the level of countries.

Static URL factor in browser logs for the maximum period. The number of leaves in URLA support. In this case, the leaves are a page from which there were no transitions. Localization to the level of countries.

Static URL factor in browser logs for the maximum period. The average time spent on the page and in all descendants of the page (URLS to which they switched) from the host. Cut if the total DT is more than 10 minutes. Localization to the level of countries.

Summary Skorov words of a request for a language model 3GRAMS-YANDEX-DIRECT.

Summary of the Skorov words of the request by the Web-Mt language model.

Static URL factor in browser logs for the maximum period. Rank, based on only UBLP meters, which allows you to find many SBR losses

The normalized amount of the scales of the words of the request that met in the text of the document or links to it.

The normalized amount of the scales of the words of the request that Equal_by_String in the text of the document or links to it.

The normalized amount of the scales of the words of the request that met in the text of the document.

The normalized amount of the scales of the words of the request that met in the links to the document.

The normalized amount of the scales of the words of the request that Equal_by_String in the links to the document.

The normalized scales for the IFILTRETRATIONMODEL words of the request that met in the text of the document or links to it.

The normalized scales for the IFILTRETRATIONMODEL Words of the request, which are Equal_by_String in the text of the document or links to it.

The normalized scales for the IFILTRETRATIONMODEL Words of the request, which Equal_by_lemma in the text of the document or links to it.

The normalized scales for the IFILTRETRATIONMODEL words of the request that met in links to the document.

The normalized scales for the IFILTRETRATIONMODEL Words of the request, which Equal_by_String in the links to the document.

Document DSSM model Language Classifier Rus.

Document DSSM model Language Classifier Eng.

Document DSSM model Language Classifier Other.

The average value of News for the year. It is calculated in offline.

ADDTIME average value for the year. It is calculated in offline.

The average value of TXTHIRELSY for the year. It is calculated in offline.

The average TEXTLIKE value is for the year. It is calculated in offline.

The average Hasnoallwordstersy value for the year. It is calculated in offline.

The average value of ISFORUM for the year. It is calculated in offline.

The average value of Haspayments for the year. It is calculated in offline.

The average value is Yabarhostavgtime2 for the year. It is calculated in offline.

The average value of Yabarurlvisitors for the year. It is calculated in offline.

The average value of QueryDowneronlyClickRate for the year. It is calculated in offline.

The average value of Dateraage for the year. It is calculated in offline.

The average value of the LonGestText for the year. It is calculated in offline.

The average value is DifferentinTernallinks for the year. It is calculated in offline.

The average value of QueryDowneronlyClickRate_Rreg is for a year. It is calculated in offline.

The average ISHUB value is for the year. It is calculated in offline.

The average value is BM25_0 on request per year. It is calculated in offline.

The average value of BOCM for the year. It is calculated in offline.

The average ISindexpage is for the year. It is calculated in offline.

The average value of queriesavgcm2 for the year. It is calculated in offline.

The average value of BrowserhostDownloadProbabolyity for the year. It is calculated in offline.

The average value of Regbrowseruserhub for the year. It is calculated in offline.

Auxtitlebm25 average value for the year. It is calculated in offline.

The average value of QuryurlCorrectedctrxFactor for the year. It is calculated in offline.

The average value is QueryTodocallsumfcountTextbM11Norm16384 for the year. It is calculated in offline.

The average value of the XFDTSHOWALLSUMWFSUMWBODYMINWIDESIZE for the year. It is calculated in offline.

Maintened by clicks ISMainPage value for the year. It is calculated in offline.

Main -heated clicks of the Yabarurlavgtime value for the year. It is calculated in offline.

Maintened by clicks DifferentinTernallinks for the year. It is calculated in offline.

Malpanized Dwelltime-AMI Value of Urldomainfraction for the year. It is calculated in offline.

BM25FDPR with standardization on the average length of the document, depending on the language of the document. Only texts are used.

Document annotations count in the whole history of the Search (DSSM AnnReg models helper)

Document annotation words count in the whole history of the Search (DSSM AnnReg models helper)

Document annotation regions count in the whole history of the Search (DSSM AnnReg models helper)

Query-MainContentKeywords similarity, target: logDwellTime

Dssm Boosting query self similarity for XfWeight model.

Dssm Boosting AvgTop02Score aggregation for XfWeight model over 5-means centroids.

Dssm Boosting AvgTop04Score aggregation for XfWeight model over 5-means centroids.

Dssm Boosting AvgTop02ScoreAvgClusterTop3Weighted aggregation for XfWeight model over 5-means centroids.

Dssm Boosting AvgTop02Score aggregation for XfWeight model over 5-means centroids (query as expansion).

Dssm Boosting AvgTop02ScoreAvgClusterTop3Weighted aggregation for XfWeight model over 5-means centroids (query as expansion).

Dssm Boosting query self similarity for XfOne model.

Dssm Boosting Score aggregation for XfOne model over 1-means centroids.

Dssm Boosting ScaledSumWeight aggregation for XfOne model over 1-means centroids.

Dssm Boosting Score aggregation for XfOne model over 1-means centroids (query as expansion).

Dssm Boosting ScoreAvgNearest1Weighted aggregation for XfOne model over 1-means centroids (query as expansion).

Dssm Boosting ScoreAvgNearest5Weighted aggregation for XfOne model over 1-means centroids (query as expansion).

Dssm Boosting Score aggregation for XfOneSe model over 1-means centroids.

Dssm Boosting ScoreScaledSumWeighted aggregation for XfOneSe model over 1-means centroids.

Dssm Boosting ScoreAvgNearest5Weighted aggregation for XfOneSe model over 1-means centroids.

Dssm Boosting query self similarity for Ctr model.

Dssm Boosting Score aggregation for Ctr model over 1-means centroids.

Dssm Boosting Score aggregation for Ctr model over 1-means centroids (query as expansion).

Dssm Boosting ScoreScaledSumWeighted aggregation for Ctr model over 1-means centroids (query as expansion).

Dssm Boosting ScoreAvgNearest1Weighted aggregation for Ctr model over 1-means centroids (query as expansion).

DSSM prediction (URL + Title), trained for the Page_QUALYY signal and implemented in RTHUB, the first slot.

DSSM prediction (URL + Title), trained on the Page_QUALYY signal and implemented in RTHUB, the second slot.

The main components of the requesting Embling from the DSSMCTRNOMINER model

The main components of the requesting Embling from the DSSMCTRNOMINER model

The main components of the requesting Embling from the DSSMCTRNOMINER model

The main components of the requesting Embling from the DSSMCTRNOMINER model

The main components of the requesting Embling from the DSSMCTRNOMINER model

The main components of the requesting Embling from the DSSMCTRNOMINER model

AVG aggregation of HasPayments web factor using random log

AVG aggregation of VideoQuery web factor using random log

AVG aggregation of SyntQuality web factor using random log

PERCENTALE_90 aggregation of GeoRegionalityVNew web factor using random log

AVG aggregation of QClassDownload web factor using random log

AVG aggregation of IsMusic web factor using random log

PERCENTALE_25 aggregation of QueryThEncyclopedic web factor using random log

AVG aggregation of CommercialOwnerRank_Reg web factor using random log

PERCENTALE_25 aggregation of YabarWordDepthNodesGradientMin web factor using random log

AVG aggregation of PopularSEFRCBrowser web factor using random log

AVG aggregation of URLClicksMaxGeoRegionFRCRatio web factor using random log

PERCENTALE_90 aggregation of UBLongPeriodDirectHChildren90CntFromExtHost web factor using random log

PERCENTALE_90 aggregation of UBLongPeriodDtUrlHChildrenCut600Reg web factor using random log

AVG aggregation of IsPicture web factor using random log

AVG aggregation of ErratumLogQueryProbability web factor using random log

Predicted by demand and country, using a DSSM model, the length of the click from this country.

The average for the year for the year predicted using the neural network.

ADDTIME ADDTIME is predicted using a neural network for a year.

The average Txthirelesy value predicted using a neural network for the year.

The average Textlike is predicted using a neural network for the year.

The average HasnoallwordStrsy is predicted using a neural network for a year.

The average ISFORUM is predicted using a neural network for the year.

The average Haspayments is predicted using a neural network for the year.

The average value of Yabarhostavgtime2 for the year for the year.

The average yabarurlvisitors is predicted using a neural network for the year.

The average value of QueryDowneronlyClickRate for the year for the year.

The average Dateraage value for the year for a year predicted using a neural network.

The average LonGestText is predicted using a neural network for the year.

The average DifferentinTernallinks for the year for the year.

The average value of QueryDowneronlyClickRate_Rreg is predicted using a neural network for a year.

The average BOCM value predicted using a neural network for the year.

The average ISindEXPAGE is predicted using a neural network for the year.

The average value of QueriesavGCM2 for the year for the year predicted using a neural network.

The average BrowserhostdowLoadProbabolyity for the year for the year.

The average value of Regbrowseruserhub for the year for a year predicted using a neural network.

The average AuxtitlebM25 average value for the year for the year.

The average value of QuryurlCorrededCTRXFACTOR for the year for the year.

The average value of QueryTodoCallsumfcountTextbM11Norm16384 for the year for the year.

The average value of the XFDTSHOWALSUMWFSUMWBODYMINWINDOWSIZE for the year for the year.

The value of the ISMAINPAGE with clicks predicted using the neural network with clicks on request for the year.

A mid Yabarurlavgtime value predicted using a neural network with clicks for a year.

DiffferentinTernallinks, which is predicted using a neural network, is a weighted net with clicks for a year.

The Malue Network DwellTime-AMI predicted using the neural network is the value of Urldomainfraction for the year.

The prediction of the probability that the request is localized in accordance with the regionality5 rule.

Dssm Boosting Score aggregation for XfOneSeAmSsHard model over 1-means centroids.

Dssm Boosting ScoreAvgClusterTop3Weighted aggregation for XfOneSeAmSsHard model over 1-means centroids.

Average by url maximum yellowness of teaser image

Average by url average yellowness of teaser image

Ratio of yellow images in teasers on host

Average yellow images count on host

Average teasers count on host

Average teasers area on host

Average by url minimum yellowness of teaser text

Average by url average yellowness of teaser text

Background is clickable advertisement

Average ratio of adverts on screen

Ratio of adverts on screen on main page

Average count of adverts on screen

Ratio of outgoing advertisement traffic to all traffic (desktop)

Ratio of outgoing real-time bidding traffic to all traffic (desktop)

Rating of news agency from agencies.json (Yandex.News resource)

Characterizes the request for the degree of change from the addition of a fixed word (number of some year), DSSM model DSSMBOOSTINGXFONESEAMSARD is used

Characterizes a request for the degree of change from the addition of a fixed word ('online' for Kirilitsa), DSSM model DSSMBOOSTINGXFONESEAMSARD is used

Characterizes the request for the degree of change from removing a fixed word ('site' for Kirilitsa), DSSM model DSSMBOOSTINGXFONESEAMSARD is used

A document from the hearts with fresh

For each word offline, the average Hasnotr meaning is calculated for 3 months. Further, in all words of the request, the maximum of this value is taken.

For each word offline, the average ISLJ value is calculated for 3 months. Further, in all words of the request, the maximum of this value is taken.

For each word offline, the average BCLMLITE value is calculated for 3 months. Further, in all words of the request, a minimum of this value is taken.

For each word offline, the average DBM40 value is calculated for 3 months. Further, for all non -feet, the words of the request are taken as a maximum of this value.

For each word offline, the average ISDESKTOPREQUEST value is calculated for 3 months. Further, for all non -feet, the words of the request are taken as a maximum of this value.

For each word offline, the average value of RLQAVGHASNOLLWORDSTRSYN is calculated at the request for 3 months. Further, in all words of the request, the maximum of this value is taken.

For each word offline, the average DSSMAGGRETEDANNREG value is calculated at the request for 3 months. Further, in all words of the request, the maximum of this value is taken.

For each word offline, the average meaning of MetanumurlSperhostfixed is calculated in demands in 3 months. Further, in all words of the request, the maximum of this value is taken.

For each word offline, the average value of MaxsdisnavmxqueryMax is calculated at the request for 3 months. Further, for all non -feet, the words of the request are taken as a maximum of this value.

AVG aggregation of VisitsFromWiki web factor using random log

PERCENTALE_25 aggregation of NavLinear web factor using random log

PERCENTALE_90 aggregation of Found web factor using random log

AVG aggregation of SubqueryThMatch web factor using random log

AVG aggregation of SegmentWordPortionFromMainContent web factor using random log

AVG aggregation of XfDtShowAllMaxFFieldSet2Bm15FLogK0001 web factor using random log

AVG aggregation of QueryRegionSize web factor using random log

The document came from Webtier1

AVG aggregation of IsRelevLocaleUA web factor using random log

PERCENTALE_90 aggregation of QfufAllSumWFSumWFieldSet3BclmWeightedFLogW0K0001 web factor using random log

PERCENTALE_90 aggregation of DssmBoostingCtrQuerySelfSimilarity web factor using random log

AVG aggregation of QueryToDocAllSumFCountTextBocm11Norm256 web factor using random log. NOTE: QueryToDocAllSumFCountTextBocm11Norm256 has been removed.

PERCENTALE_90 aggregation of IsNavMxQuery web factor using random log

The document came from Platinum0

AVG aggregation of DBM15Wares2 web factor using random log

PERCENTALE_90 aggregation of UrlNGramsModel web factor using random log

PERCENTALE_25 aggregation of DssmBoostingCtrKMeans1ScoreScaledSumWeightedQE web factor using random log

PERCENTALE_90 aggregation of LongClickMobileAllWcmWeightedValue web factor using random log

PERCENTALE_25 aggregation of DssmVkPopularity web factor using random log

AVG aggregation of UBLongPeriodVisitsSNProb web factor using random log

PERCENTALE_90 aggregation of CountryQueryRegionality web factor using random log

PERCENTALE_90 aggregation of TRhitw web factor using random log

PERCENTALE_90 aggregation of UBLongPeriodAvgSearchDuration600 web factor using random log

AVG aggregation of RequestIsFromIOS web factor using random log

PERCENTALE_90 aggregation of DssmQueryEmbeddingCtrNoMinerPca4 web factor using random log

AVG aggregation of XfDtShowAllMaxFFieldSetUTBm15FLogW0 web factor using random log

PERCENTALE_25 aggregation of UrlTrigrams web factor using random log

PERCENTALE_90 aggregation of DssmQueryEmbeddingCtrNoMinerPca1 web factor using random log

AVG aggregation of IsRelevLocaleKZ web factor using random log

PERCENTALE_90 aggregation of TextFeatures web factor using random log

1 if host include js from marketgid.com

1 if host include js from rfity.com

DSSM prediction of google specificity for query

1 if host include js from google-analytics.com

1 if host include js from googleapis.com

1 if host include js from facebook.net

1 if host include js from mc.yandex.ru

The average value of Randomlogqueryavgaddtime of the closest KNN queries.

The average value of RandomlogqueryavgtXthirelsy nearest KNN queries.

The average value of Randomlogqueryavgtextlike nearest KNN queries.

The average value of Randomlogqueryavgisforum of the closest KNN queries.

The average value of Randomlogqueryavghaspayments closest to KNN queries.

The average value of Randomlogqueryavgdiferentinternallinks of the nearest KNN queries.

The average value of RandomlogqueryavgistargetbussinessCard of the nearest KNN queries.

The average value is RandomlogqueryavgquerytododoCallsumfcountTextBM11NORM16384 of the nearest KNN queries.

The average value is Randomlogqueryavgxfdtshowallsumwfsumwbodyminwindowsize closest KNN queries.

Host speed estimation

Is site official

Quality link from good sites estimation

Weight sum of each non-unique nevasca shingle

Nevasca shingle quantity in last week

Greentraffic share (aka direct visits). Desktop

Greentraffic share (aka direct visits). Mobile

Greentraffic absolute (desktop)

Visits averaged by user

1 if video on page

RMSE aggregation of Long web factor using random log

RMSE aggregation of IsOrg web factor using random log

RMSE aggregation of GskUrlModel web factor using random log

RMSE aggregation of DaterStatsAverageSourceSegment web factor using random log

RMSE aggregation of VisitsFromWiki web factor using random log

RMSE aggregation of XfDtShowBagOfWordsTitleCosineMaxMatch web factor using random log

RMSE aggregation of UBLongPeriodDownloadsProb web factor using random log

RMSE aggregation of MetaAvgIsNotCgi meta factor using random log

RMSE aggregation of MetaRmsSynPercentBadWordPairs meta factor using random log

RMSE aggregation of MetaPosTrigramsProb meta factor using random log

PERCENTALE_90 aggregation of Bocm web factor using random log

PERCENTALE_90 aggregation of SegmentWordPortionFromMainContent web factor using random log

PERCENTALE_90 aggregation of IsMobileBeauty web factor using random log

PERCENTALE_90 aggregation of USLongPeriodUrlWinsProb web factor using random log

PERCENTALE_90 aggregation of DssmBoostingXfWeightKMeans5AvgTop02ScoreQE web factor using random log

PERCENTALE_90 aggregation of DssmBoostingCtrKMeans1Score web factor using random log

PERCENTALE_90 aggregation of SDIsNavMxQueryMax meta factor using random log

PERCENTALE_90 aggregation of MetaWeb764Web1076ProductInvAvg meta factor using random log

PERCENTALE_90 aggregation of MetaWeb1099Web1219ProductInvPos meta factor using random log

PERCENTALE_90 aggregation of MetaMaxDssmMiddleVsShortLongHardNoClicks meta factor using random log

MAX aggregation of NumLinksFromMP web factor using random log

MAX aggregation of NavLinear web factor using random log

MAX aggregation of DaterStatsAverageSourceSegment web factor using random log

MAX aggregation of WeightedSumIsIndexPageIsNavMxQuery web factor using random log

MAX aggregation of QueryToDocAllSumFCountTextBocm11Norm256 web factor using random log. NOTE: QueryToDocAllSumFCountTextBocm11Norm256 has been removed.

MAX aggregation of DssmBigramsQueryDerivativeMax web factor using random log

MAX aggregation of DssmQueryCountryToUrlEstimatedDistance web factor using random log

MAX aggregation of MetaWeb764Web1076ProductInvAvg meta factor using random log

LOGAVG aggregation of TextFeatures web factor using random log

LOGAVG aggregation of DocLen web factor using random log

LOGAVG aggregation of IsHTML web factor using random log

LOGAVG aggregation of HasLevensht1QueryFragment web factor using random log

LOGAVG aggregation of HeadingIdfSumFixed web factor using random log

LOGAVG aggregation of AdvPronounsPortion web factor using random log

LOGAVG aggregation of LongestText web factor using random log

LOGAVG aggregation of CountryHour web factor using random log

LOGAVG aggregation of MetrikaUrlAvgTime web factor using random log

LOGAVG aggregation of WikiLinkCount web factor using random log

LOGAVG aggregation of BrowserUrlDwellTimeRegionFrc web factor using random log

LOGAVG aggregation of WikiInfobox web factor using random log

LOGAVG aggregation of QueryDocTitleRangesMatchingScore web factor using random log

LOGAVG aggregation of IsMobileBeauty web factor using random log

LOGAVG aggregation of QueryToTextAllSumWFSumWBodyMinWindowSize web factor using random log

LOGAVG aggregation of DssmRandomLogQueryAvgDifferentInternalLinks web factor using random log

LOGAVG aggregation of MetaUrlDirectChildrenCnt meta factor using random log

LOGAVG aggregation of MetaWeb1241Web1299ProductInvPos meta factor using random log

LOGAVG aggregation of MetaEpsHashShareNationalLanguage meta factor using random log

The document has a HTTPS protocol

Levenstein’s distance between the request and Url of the type of YouTubecom/Watch normalized to the maximum of the length of the request and Url

The length of the largest total setting between Url and the request normalized for the length of the request

Sigmoid rationed the value of a textual classifier of porn according to Toloka Porn

Binarized value of a textual classifier text classifier according to Toloka estimates

The value of the text classifier of porn according to the classifier of the web and add. dictionaries

Binarized with the use of networks, the value of a textual classifier of porn according to the estimates of the web and additional classifier. dictionaries

The presence of obscene vocabulary in the request. 0 - absent, 0.5 - non -seated, 1 - hard

The presence of porn markers in the request (0 - is, 1/3 - no, 1 - request 'gray')

Documentary classifier of porn, features according to the text of the document

Documentary classifier of porn, features on Ural document

Documentary classifier of porn, features according to the pictures of the document (information is taken from the picture index)

Documentary classifier of porn, features by video of the document (information is taken from the video index)

A host classifier of porn, features about the porn of the requests, according to which the host was shown and clung.

The presence in a lemmetized request to buy words, price and similar in meaning

Return factor to the host. Percentale aggregation with a coefficient of 0.25F Fichi Dwelltimesumfration

The document came from Quickmed

Return factor to the host. Percentale aggregation with a coefficient of 0.99f Fichi AveraGereturntime

Return factor to the host. Percentale aggregation with a coefficient of 0.97f feature AveraGereturntime

Return factor to the host. Greaterfraction Aggregation with a coefficient of 0.99F Fichi Averageturntime

Return factor to the host. Percentale aggregation with a coefficient of 0.99F Fichi AveraGelogreturntime

Return factor to the host. Greaterfraction Aggregation with a coefficient of 0.9F Fichi AveraGelogreturntime

Return factor to the host. Lessfraction Aggregation with a coefficient of 0.05F FIRSTCLickdwelltime Fichi

Host factor. Weigtedaverage aggregation Fichi averagevisitper3hors

Host factor. Weigtedaverage aggregation Fichi AvegedwellTimeperhour

Return factor to the host. Lessfraction Aggregation with a coefficient of 0.1F Fichi Averagewelltimeper3Hows

Host factor. Max aggregation Fichi AvegedwellTimeperweek

Median dwelltai request in history. Dwelltaym is cut to 6000. The request is normalized by doppelgangers

The number of shows of the request with more than one click in history. The request is normalized by doppelgangers

The share of shows with more than one click from all shows in history. The request is normalized by doppelgangers

Owner aggregation of RandomLogWordMaxMetaNumUrlsPerHostFixed web factor using random log, aggregation type is PERCENTALE_90

Owner aggregation of MetaWeb1099Web1219ProductInvPos meta factor using random log, aggregation type is LOGAVG

Owner aggregation of DssmDwelltimeRegChainTrainedEmbedding meta factor using random log, aggregation type is PERCENTALE_90

Owner aggregation of DssmRandomLogQueryAvgHasPayments web factor using random log, aggregation type is LOGAVG

Owner aggregation of UBLongPeriodBrowseFrc web factor using random log, aggregation type is PERCENTALE_90

Owner aggregation of MetaUrlChildrenCnt meta factor using random log, aggregation type is LOGAVG

Owner aggregation of MetaRmsDifferentInternalLinks meta factor using random log, aggregation type is PERCENTALE_25

Owner aggregation of RandomLogWordMaxHasNoTr web factor using random log, aggregation type is PERCENTALE_90

Owner aggregation of MetaResidUSLongPeriodUrlWinsProb meta factor using random log, aggregation type is RMSE

Owner aggregation of PornoQuery web factor using random log, aggregation type is LOGAVG

Owner aggregation of NationalLanguage web factor using random log, aggregation type is LOGAVG