CN111597449A

CN111597449A - Candidate word construction method and device for search, electronic equipment and readable medium

Info

Publication number: CN111597449A
Application number: CN202010431955.8A
Authority: CN
Inventors: 彭睿棋
Original assignee: Beijing ByteDance Network Technology Co Ltd
Current assignee: Beijing ByteDance Network Technology Co Ltd
Priority date: 2020-05-20
Filing date: 2020-05-20
Publication date: 2020-08-28
Anticipated expiration: 2040-05-20
Also published as: CN111597449B

Abstract

The disclosure provides a candidate word construction method and device for searching, electronic equipment and a readable medium, and relates to the technical field of computers. The method comprises the following steps: determining a plurality of aged search terms associated with time based on historical search data; acquiring structural information corresponding to the aging search word, and generating a plurality of aging candidate words based on the aging search word and the structural information corresponding to the aging search word; the structured information comprises a plurality of entity information formed by extracting information from webpage data; and constructing a candidate word set of the target user based on the plurality of aging candidate words and the characteristic information of the target user. The aging candidate words are generated based on the structural information corresponding to the aging search words and the aging search times, and the candidate word set of the target user is constructed based on the aging candidate words and the characteristic information of the target user, so that the scale of the candidate words which can be recommended to the target user is greatly improved.

Description

Candidate word construction method and device for search, electronic equipment and readable medium

Technical Field

The present disclosure relates to the field of computer technologies, and in particular, to a candidate word construction method and apparatus for search, an electronic device, and a readable medium.

Background

In the search engine, when a user initiates a search in a search box, some personalized candidate words are recommended to the user through the search box, the user can use the recommended candidate words as search words for searching, the process reduces the search cost of the user by guessing the search intention of the user, and meanwhile, the search requirement of the user is stimulated to a certain extent.

In the prior art, when recommending candidate words to a user, at least one search word is generally selected from search words in a search engine database as a candidate word for recommendation, and the search words in the search engine database are generally historical search words input by the user and have a small scale. Before the search words in the search engine database are used as candidate words to be recommended to the user, the search words in the search engine database need to be screened, and the search words meeting the conditions can be recommended to the user as the candidate words; limited by the scale of historical search terms stored in the database and the screening efficiency, candidate terms which can be recommended to the user are very limited, and the personalized search requirements of the user are difficult to meet, and the search requirements of the user on timeliness cannot be met.

Disclosure of Invention

This summary is provided to introduce a selection of concepts in a simplified form that are further described below in the detailed description. This summary is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used to limit the scope of the claimed subject matter.

In a first aspect of the present disclosure, a candidate word construction method for search is provided, including: determining a plurality of aged search terms associated with time based on historical search data; acquiring structural information corresponding to the aging search word, and generating a plurality of aging candidate words based on the aging search word and the structural information corresponding to the aging search word; the structured information comprises a plurality of entity information formed by extracting information from webpage data; and constructing a candidate word set of the target user based on the plurality of aging candidate words and the characteristic information of the target user.

In a second aspect of the present disclosure, a candidate word constructing apparatus for searching is provided, including: a determination module to determine a plurality of aged search terms associated with time based on historical search data; the generating module is used for acquiring structural information corresponding to the aging search words and generating a plurality of aging candidate words based on the aging search words and the structural information corresponding to the aging search words; the structured information comprises a plurality of entity information formed by extracting information from webpage data; and the constructing module is used for constructing a candidate word set of the target user based on the plurality of aging candidate words and the characteristic information of the target user.

In a third aspect of the present disclosure, an electronic device is provided, which includes: a memory and a processor; the memory has a computer program stored therein; a processor for performing the method of the first aspect when executing the computer program.

In a fourth aspect of the disclosure, a computer-readable medium is provided, on which a computer program is stored, which computer program, when being executed by a processor, is adapted to carry out the method of the first aspect.

The technical scheme provided by the disclosure has the following beneficial effects:

the method and the device for searching the target user candidate words generate the plurality of the aging candidate words based on the aging search words and the structural information corresponding to the aging search words, and construct the candidate word set of the target user based on the plurality of the aging candidate words and the characteristic information of the target user, so that the scale of the candidate words which can be recommended to the user is greatly improved, and the problem that the number of the candidate words which can be recommended to the user is too small due to the fact that the screening efficiency of screening the search words stored in a search engine database is low is solved. The method and the device improve the scale of the candidate words which can be recommended to the user, and are beneficial to meeting the personalized search requirement of the user; and furthermore, the timeliness of the timeliness search word is given to the timeliness candidate word, so that the search requirements of the user on timeliness are favorably met when the timeliness candidate word is recommended to the user, and the search experience of the user is improved.

Drawings

The above and other features, advantages and aspects of various embodiments of the present disclosure will become more apparent by referring to the following detailed description when taken in conjunction with the accompanying drawings. Throughout the drawings, the same or similar reference numbers refer to the same or similar elements. It should be understood that the drawings are schematic and that elements and features are not necessarily drawn to scale.

Fig. 1 is a flowchart of a first candidate word construction method for searching according to an embodiment of the present disclosure;

fig. 2 is a flowchart of a second candidate word construction method for searching according to an embodiment of the present disclosure;

fig. 3 is a flowchart of a third candidate word construction method for searching according to an embodiment of the present disclosure;

fig. 4 is a flowchart of a fourth candidate word construction method for searching according to an embodiment of the present disclosure;

fig. 5 is a flowchart of a fifth candidate word construction method for searching according to an embodiment of the present disclosure;

fig. 6 is a flowchart of a sixth candidate word construction method for searching according to an embodiment of the present disclosure;

fig. 7 is a flowchart of a seventh candidate word construction method for searching according to an embodiment of the present disclosure;

fig. 8 is a flowchart of an eighth candidate word construction method for searching according to an embodiment of the present disclosure;

fig. 9 is a flowchart of a ninth candidate word construction method for searching according to an embodiment of the present disclosure;

fig. 10 is a schematic structural diagram of a candidate word constructing apparatus for searching according to an embodiment of the present disclosure;

fig. 11 is a schematic structural diagram of an electronic device according to an embodiment of the present disclosure.

Detailed Description

Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While certain embodiments of the present disclosure are shown in the drawings, it is to be understood that the present disclosure may be embodied in various forms and should not be construed as limited to the embodiments set forth herein, but rather are provided for a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the disclosure are for illustration purposes only and are not intended to limit the scope of the disclosure.

It should be understood that the various steps recited in the method embodiments of the present disclosure may be performed in a different order, and/or performed in parallel. Moreover, method embodiments may include additional steps and/or omit performing the illustrated steps. The scope of the present disclosure is not limited in this respect.

The term "include" and variations thereof as used herein are open-ended, i.e., "including but not limited to". The term "based on" is "based, at least in part, on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Relevant definitions for other terms will be given in the following description.

It should be noted that the terms "first", "second", and the like in the present disclosure are only used for distinguishing the devices, modules or units, and are not used for limiting the devices, modules or units to be different devices, modules or units, and also for limiting the sequence or interdependence relationship of the functions executed by the devices, modules or units.

It is noted that references to "a", "an", and "the" modifications in this disclosure are intended to be illustrative rather than limiting, and that those skilled in the art will recognize that "one or more" may be used unless the context clearly dictates otherwise.

The names of messages or information exchanged between devices in the embodiments of the present disclosure are for illustrative purposes only, and are not intended to limit the scope of the messages or information.

To make the objects, technical solutions and advantages of the present disclosure more apparent, embodiments of the present disclosure will be described in detail with reference to the accompanying drawings.

The disclosure will be described and explained with reference to several terms:

historical search data: historical search data characterizes a series of activities of a user to search, make decisions, etc. through a search engine. Data such as a webpage accessed by a user, a clicked result, a used search word query, a search sequence and the like can be extracted or obtained from historical search data; the historical search data is search data of all users using the search engine currently stored in a database of the search engine, and is continuously updated along with the use of the search engine by the users, and the historical search data can also be understood as a search log.

Structuring information: the structured information means that the information can be decomposed into a plurality of components which are mutually related after being analyzed, and each component has a clear hierarchical structure, the use and maintenance of the hierarchical structure are managed through a database, and certain operation specifications are provided. Structured information is information that is typically provided by multiple partners and that can be represented as information that a search engine recommends multiple entities involved in each page in a search result based on search requirements. If based on shopping search, after a webpage is captured by a search engine, commodity information in the webpage is extracted, information such as commodity names, prices, brief introduction and the like is extracted, even commodities such as notebook brief introduction can be further subdivided into 'brands, models, CPUs, memories, hard disks, display screens' and the like, the extracted information is integrated into structured information, and subsequently when the search requirement is shopping, relevant information of each entity can be known based on the structured information. In the embodiment of the present disclosure, the structured information adopts information existing in the database, such as a plurality of entity information formed by extracting information from the captured web page based on the historical search data.

And (3) aging search words: refers to search terms with timeliness, and is determined based on demand mining on historical search data. Wherein, the timeliness at least comprises three types: long timeliness, periodic timeliness, burst timeliness; the search words with long timeliness exist continuously, and the frequency change of the search words is not large, such as air temperature; the periodic aging search words have periodicity, and the frequency of the occurrence of the periodic aging search words changes periodically, such as fourth and sixth levels and TV series; the emergent time search word has an emergent property, and the frequency of the emergent time search word changes along with the occurrence time of an emergent event, such as an earthquake, a sports event and the like.

And (4) aging candidate words: generating structural information corresponding to the aging search words based on the aging search words, wherein the aging candidate words also belong to the candidate words with the aging property corresponding to the aging property of the aging search words; such as: performing demand mining in historical search data, and determining a search word with periodic timeliness, such as the search word 'ABC TV play'; and generating an aging candidate word such as an X-th episode of the ABC TV play based on the aging search word and the corresponding structural information (information of the network video playing platform about the update time, the update episode, the watching permission and the like of the ABC TV play).

The life cycle is as follows: specifically, the life cycle of the aging candidate word is generally several hours or several days, for example, the aging candidate word "ground trap" with sudden aging is adapted to the time for the relevant news platform to update the news broadcast, for example, 12 hours, the life cycle of the aging candidate word is set to 12 hours, and the 12 hours are ended, and the aging candidate word is invalid; if the aging candidate word with periodic aging is the 1 st set of the XXX television series, adapting the network video playing platform to update the update time of the television series, if updating every saturday, setting the life cycle of the aging candidate word to be one week, after one week, ending the life cycle of the aging candidate word of the 1 st set of the XXX television series, and accordingly, failing to be the aging candidate word.

The following describes the technical solutions of the present disclosure and how to solve the above technical problems in specific embodiments. The following several specific embodiments may be combined with each other, and details of the same or similar concepts or processes may not be repeated in some embodiments. Embodiments of the present disclosure will be described below with reference to the accompanying drawings.

Referring to fig. 1, the present disclosure provides a candidate word construction method for search, which may be specifically executed by an electronic device according to an embodiment of the present disclosure, and specifically, the electronic device may be a server, where the present disclosure includes:

s101, determining a plurality of aging search words associated with time based on historical search data;

s102, acquiring structural information corresponding to the aging search word, and generating a plurality of aging candidate words based on the aging search word and the structural information corresponding to the aging search word; the structured information comprises a plurality of entity information formed by extracting information from webpage data;

s103, constructing a candidate word set of the target user based on the plurality of aging candidate words and the characteristic information of the target user.

In step S101, a plurality of aging search terms associated with time are determined based on historical search data; specifically, historical search data currently stored in a database of a search engine is subjected to demand mining, search demands are mined, timeliness is mined at the same time, and a plurality of timeliness search terms associated with time are determined. The timeliness comprises long timeliness, periodic timeliness and burst timeliness, and can be determined through search behavior data corresponding to historical search words in historical search data.

For example, the aging search term with long aging includes search terms which exist continuously in the historical search data, such as "air temperature"; the periodically-aged searching words comprise periodically-appeared searching words in historical searching data, such as TV drama and level four and six tests; the search terms of the sudden timeliness comprise search terms which appear suddenly in historical search data, such as 'earthquake' and 'CBA event'.

Optionally, the database of the search engine includes historical search data of all users using the search engine; in the embodiment of the present disclosure, when performing demand mining, the adopted historical search data may be historical search data of some users using a search engine, such as historical search data of users whose usage frequency reaches a preset frequency threshold; or historical search data of all users using the search engine.

In step S102, obtaining structural information corresponding to the aging search term, and generating a plurality of aging candidate terms based on the aging search term and the structural information corresponding to the aging search term; the structured information comprises a plurality of entity information formed by extracting information from webpage data. Specifically, structural information stored in a database is obtained, and an aging search word is mapped to corresponding structural information to obtain structural information corresponding to the aging search word; in mapping, the time effect search word can be analyzed through semantic analysis, attribute analysis and the like, query is carried out in a database according to an analysis result, and the structural information related to the time effect search word is determined as the structural information corresponding to the time effect search word. And after the structural information corresponding to the aging search word is obtained, combining the aging search word and the structural information to generate a plurality of aging candidate words. Optionally, the aging search term is directly matched with the structural information stored in the database, and if the aging search term is matched with the structural information stored in the database, the related information is acquired as the structural information corresponding to the aging search term.

For example, the following steps are carried out: when the aging search word is a search word "drama" with periodic aging, structural information related to the drama is inquired in a database storing a plurality of pieces of structural information after the aging search word is analyzed, and if the structural information comprises 100 dramas and related episodes, corresponding aging candidate words such as "drama XXX 1 st episode" and "drama yyyy 3 rd episode" can be generated by combining the aging search word. When the aging search word is a search word 'earthquake' with paroxysmal aging, structural information related to the earthquake is inquired in a database storing a plurality of structural information after the aging search word is analyzed, news event keywords related to the earthquake in the latest time period, such as place name, time, magnitude and the like, are included in the structural information, and corresponding aging candidate words, such as '6.6-level earthquake' and '512 earthquake', are generated by combining the aging search word. When the aging search word is a search word 'car' with long aging, the aging search word is analyzed to determine that the search word is related to the car, structured information related to the car is inquired in a database in which a plurality of pieces of structured information are stored, the structured information comprises 1000 kinds of cars with different brands, and the corresponding attributes of the 1000 kinds of cars are 100 kinds, so that 10 ten thousand aging candidate words can be generated by combining the aging search word.

In step S103, constructing a candidate word set of the target user based on the plurality of aging candidate words and feature information of the target user; specifically, the aging candidate words generated in step S102 are used as candidate words of the whole network, and when the candidate words are adapted to a single target user, at least one aging candidate word that can be recommended to the target user is determined by calculating the similarity between the plurality of aging candidate words and the feature information of the target user, and a candidate word set of the target user is formed. The similarity may be calculated by using cosine similarity, euclidean distance, weighted minkowski distance, or other methods to calculate the feature information of the aging candidate word and the target user, which are not exhaustive here.

In an embodiment, a preset recommendation model is adopted to score and sort a plurality of aging candidate words so as to construct a candidate word set of a target user. In the ranking stage of the recommendation model, if any aging candidate word enters the optimal decision result after model ranking, the probability of adoption by the target user is higher when the target user recommends the aging candidate word as the candidate word. Therefore, any aging candidate word is determined to be a candidate word which can be recommended to the target user. When a target user initiates a search request, recommending candidate words in the candidate word set to the target user, or recommending at least one corresponding aging candidate word in the candidate word set of the target user to the user based on the search request. Optionally, the recommendation model may be established by using one of an FM model, a GBDT + LR model, a DNN model, and the like, and is mainly used for accurately sorting the time-efficient candidate words in the sorting stage.

Alternatively, the generation process of the aging candidate words is a continuous activity, the historical search data stored in the database is continuously updated with the data generated when the user searches through the search engine, accordingly, step S101 identifies more aging search words associated with time, and step S102 continuously generates new aging candidate words based on the continuously updated aging search words and the corresponding structural information. Optionally, the aging search word with periodic aging may be combined with the time information in the corresponding structural information to form a plurality of aging candidate words without interruption in each period, and if an ABC series is updated six weeks each, the aging candidate word of the "xth episode of the ABC series" is generated six weeks each in combination with the episode information of the ABC series update until the series update is finished.

The method and the device for searching the target user candidate words generate the plurality of the aging candidate words based on the aging search words and the structural information corresponding to the aging search words, and construct the candidate word set of the target user based on the plurality of the aging candidate words and the characteristic information of the target user, so that the scale of the candidate words which can be recommended to the user is greatly improved, and the problem that the number of the candidate words which can be recommended to the user is too small due to the limitation of the screening efficiency of screening the search words stored in a search engine database is solved. The method and the device improve the scale of the candidate words which can be recommended to the user, and are beneficial to meeting the personalized search requirement of the user; and furthermore, the timeliness of the timeliness search word is given to the timeliness candidate word, so that the search requirements of the user on timeliness are favorably met when the timeliness candidate word is recommended to the user, and the search experience of the user is improved.

In one embodiment, referring to fig. 2, step S101 determines, based on historical search data, a plurality of age search terms associated with time, including at least one of:

s201, obtaining multiple times of search behavior data in a preset time interval from historical search data, and if the number of the search behavior data aiming at the same historical search word in the preset time interval exceeds a first preset number threshold, extracting a keyword from the search behavior data corresponding to the historical search word to be used as a sudden aging search word.

S202, acquiring multiple times of search behavior data in a preset time interval from historical search data, and if any historical search word is identified to be in accordance with the preset paroxysmal timeliness characteristics and the number of the search behavior data aiming at any historical search word in the preset time interval exceeds a second preset number threshold, extracting a keyword from the search behavior data corresponding to any historical search word to be used as a paroxysmal timeliness search word.

S203, if the periodic search behavior data aiming at the same historical search word is extracted from the historical search data, and the number of the search behavior data in each period exceeds a third preset number threshold, extracting the keywords from the search behavior data corresponding to the historical search word as periodic aging search words.

Specifically, the search behavior data includes historical search terms, a plurality of search results corresponding to the historical search terms, and a page corresponding to at least one search result clicked and/or browsed by the user. The preset time interval may be set to the latest period of time, such as the latest day; in addition, considering that the generation of the aging candidate word in the embodiment of the present disclosure belongs to a persistent activity, the preset time interval may also be determined based on the frequency of using the search engine by the user, and if the number of times of using a month reaches a preset threshold, the preset time interval is set to be the latest month.

In step S201, determining whether the number of search behavior data of a historical search word in a preset time interval exceeds a first preset number threshold is used as a condition for determining whether the historical search word has bursty aging, and when the condition is met, determining that the historical search word has bursty aging, and further extracting a keyword from the search behavior data corresponding to the historical search word as a bursty aging search word. The number of search behavior data of the historical search terms may be understood as the number of pages clicked and/or browsed by the user in the search results corresponding to the historical search terms. And extracting keywords from the search behavior data corresponding to the historical search terms, namely extracting keywords from the search results corresponding to the historical search terms and pages corresponding to at least one search result clicked and/or browsed by the user. For example, when it is determined that the historical search term "tiger-door bridge" has a sudden aging, keywords are extracted from the search behavior data corresponding to the historical search term "tiger-door bridge", and words such as "water horse", "vibration", "recovery traffic", and "tiger-door bridge" are used as the sudden aging search term.

In step S202, it is determined that the historical search word has bursty aging and two conditions need to be satisfied, where in the first condition, the historical search word conforms to a predetermined bursty aging characteristic; and a second condition that the quantity of the search behavior data for the historical search term in a preset time interval exceeds a second preset quantity threshold. In the first condition, the sudden aging characteristic can be understood as that the historical search words comprise preset characteristic words, and the preset characteristic words comprise words such as 'latest', 'latest message', 'recent' and the like; and if the historical search word 'the latest message of the Tiger gate bridge' comprises the preset feature word 'the latest message', determining that the historical search word 'the latest message of the Tiger gate bridge' meets a first condition. The second condition is the same as step S201, and the first preset number threshold and the second preset number threshold may be set to be the same threshold, or may also be set to be different thresholds. And when the current judged historical search word has the sudden aging, extracting the key word from the search behavior data corresponding to the historical search word as a sudden aging search word.

Optionally, in an embodiment, the historical search terms may also be judged by solely using the first condition, and when the first condition is satisfied, it is determined that the currently judged historical search terms have a sudden aging.

In step S203, extracting periodic search behavior data for the same historical search word from the historical search data, i.e. the search behavior data extracted in step S203 has periodicity, which is different from step S201 and step S202; for example, periodic search behavior data of the historical search term "college entrance examination" is extracted. Further, when judging whether the historical search term "college entrance examination" has periodic timeliness, the number of search behavior data in each period is required to exceed a third preset number threshold, that is, the peak number of search behavior data in each period exceeds the third preset number threshold; for example, in the periodic search behavior data of the historical search word "college entrance examination", the number of the search behavior data corresponding to the historical search word "college entrance examination" reaches the maximum value in 5, 6 and 7 months in each year, and the maximum value is compared with the third preset number threshold value to judge whether the third preset number threshold value is exceeded. And when the current judged historical search word is determined to have periodic aging, extracting keywords from the search behavior data corresponding to the historical search word as periodic aging search words. For example, the keywords "college entrance examination time", "college entrance examination question", etc. are extracted from the search behavior data corresponding to the historical search term "college entrance examination" as the periodic aging search terms.

In an embodiment, referring to fig. 3, step S102 obtains structural information corresponding to the aging search term, where the structural information includes at least one of the following items:

s301, according to the periodic aging search word, inquiring from a database in which a plurality of pieces of structural information are stored, and determining the structural information containing the periodic aging search word as the structural information corresponding to the aging search word;

s302, according to the sudden aging search word, inquiring from a database in which a plurality of structural information are stored, and determining the structural information containing the sudden aging search word as the structural information corresponding to the aging search word;

s303, according to the sudden aging search term and hot query topic information associated with the sudden aging search term, querying from a database storing a plurality of structural information, and determining partial structural information which is related to the hot query topic information and contains the structural information of the sudden aging search term as structural information corresponding to the aging search term;

the structured information comprises an entity information subject, a plurality of attributes corresponding to the entity information subject and entity keywords included by each attribute.

In step S301, processing may be performed based on the periodic aging search term determined in step S203; specifically, a query is made from a database storing a plurality of pieces of structural information according to the periodic aging search term, and the structural information containing the periodic aging search term is determined as the structural information corresponding to the aging search term. For example, if the periodic aging search term is "college entrance examination", the database is queried based on the periodic aging search term "college entrance examination", and a plurality of pieces of structured information related to "college entrance examination" are obtained and determined as structured information corresponding to the aging search term, for example, a plurality of entity information topics, a plurality of attributes corresponding to the entity information topics, and an entity keyword included in each attribute are stored in the database in a table form, and if the database is queried by using "college entrance examination" as a query term, a plurality of attributes corresponding to the entity information topics and an entity keyword included in each attribute can be obtained, and the following description is provided in conjunction with table 1:

TABLE 1

As can be seen from table 1 above, querying the database with the periodic aging search term "college entrance examination" can obtain a plurality of attributes, such as "time", "place", "subject", and the like, corresponding to the entity information subject with "college entrance examination" and the entity keyword included in each attribute.

In step S302, processing may be performed based on the bursty aging search term determined in step S201 and/or step S202; specifically, according to a sudden aging search word, a query is made from a database in which a plurality of pieces of structural information are stored, and the structural information including the sudden aging search word is determined as the structural information corresponding to the aging search word. The difference between step S302 and step S301 is that step S301 performs processing based on the periodic aging search term, step S302 performs processing based on the burst aging search term, and the processing manners of querying and obtaining corresponding structured information in the database are the same, which are not described herein again.

In step S303, processing may be performed based on the bursty aging search term determined in step S201 and/or step S202; specifically, according to the sudden aging search term and hot query topic information associated with the sudden fourteen-small search terms, query is performed from a database in which a plurality of pieces of structural information are stored, and partial structural information in the structural information containing the sudden aging search term and associated with the hot query topic information is determined as decoupling thigh information corresponding to the aging search term. The structural information A corresponding to the sudden aging search term can be inquired in a database based on the sudden aging search term, then the structural information B related to the hot inquiry topic information is obtained in the structural information A based on the hot inquiry topic information, and the structural information B is used as the structural information corresponding to the aging search term. Alternatively, step S303 may also directly perform query in the database based on the paroxysmal aging search term and the trending query topic information to obtain the structured information C, and use the structured information C as the structured information corresponding to the aging search term. The topical query topic information may be understood as information viewed by the user and/or a representation of a clicked web page in a search result corresponding to the paroxysmal aging search term, for example, a keyword of the web page clicked by the user in the search result is taken as the topical query topic information.

The plurality of pieces of structural information stored in the database can be formed by extracting information of the captured webpage data based on historical search data. Further, the structured information may also be formed by information provided by a plurality of partners, which may be directly stored based on the structured information provided by the partners, or the structured information applied to the search engine may be constructed based on the network structure of the search engine based on the data provided by the partners. Specifically, if the partner is a car vendor that provides data that includes 1000 different types of cars and 100 attributes associated with the cars, it can structure the data, record the data provided by the partner in a digitized form, and store it in a database that forms part of the structured information.

In an embodiment, referring to fig. 4, the generating a plurality of aging candidate words based on the aging search word and the structural information corresponding to the aging search word in step S102 includes at least one of:

s401, generating a plurality of aging candidate words based on the sudden aging search words, each entity keyword included in the structural information corresponding to the sudden aging search words and a preset lexical item of sudden aging;

s402, generating a plurality of aging candidate words based on the periodic aging search words and each entity keyword included in the structural information corresponding to the periodic aging search words.

In step S401, a plurality of aging candidate words are generated based on the sudden aging search word in combination with the preset terms of the sudden aging and the entity keywords included in the structural information corresponding to the sudden aging search word; specifically, search terms based on bursty aging, such as "Sichuan earthquake"; inquiring in a database to obtain entity keywords included in structured information including Ganzui, Shiduxian, Yongyuan, Rongxian, longitude and latitude (northern latitude XX.XX, east longitude XX.XX) and the like; because the aging search word has sudden aging, through big data analysis, the preset terms can be set as words such as 'latest message', 'emergency' and the like, and a plurality of aging candidate words such as 'latest earthquake message in Shichen county, Sichuan Mingqin county, latest earthquake message in Rongcounty, Sichuan Mingqin county, latest earthquake message in Sichuan Mingqin county' and the like can be generated through combination.

In step S402, generating a plurality of aging candidate words in a combined manner based on the periodic aging search word and each entity keyword included in the corresponding structural information; inquiring in a database to obtain a plurality of entity keywords included in structured information including text synthesis, theory synthesis, Jiangxi province and the like based on a periodic time search word such as 'college entrance examination'; the combination may generate a plurality of aging candidate words such as "high-school entrance and study", "Jiangxi province high-school entrance and study", and the like.

Optionally, the generation time of the aging candidate word may also be set based on the aging of the periodic aging search word. For example, the generation time of the aging candidate word is set based on the determination time of the periodic aging search word (time of mining the demand), or the generation time of the aging candidate word is determined based on the aging of the periodic aging search word. For example, the aging candidate word "tv series XXX set X" belongs to an aging candidate word with periodic aging, and may be generated based on its update time or mapping time, such as generating an aging candidate word "tv series XXX set 2" on the day of mapping the 2 nd set, and generating an aging candidate word "tv series XXX set 3" on the day of mapping the 3 rd set.

In an embodiment, if the search term further includes a long-timeliness search term, generating a plurality of timeliness candidate terms based on the long-timeliness search term by combining a plurality of entity keywords included in the structured information; in particular, based on long-term timeliness search terms, such as "car"; inquiring in a database to obtain a plurality of entity keywords included in the structural information of the public, Audi, Toyota, dynamic property, fuel economy, braking property, control stability and the like; the combination may generate a plurality of age candidates such as "mass cars", "toyota car handling stability", "audi car dynamics", and the like.

In one embodiment, there are multiple aging search terms associated with time derived based on historical search data, which may also be classified into different categories, respectively. Different aging search words may generate the same or similar aging candidate words when combined with a plurality of entity keywords included in corresponding structured information, or the same or similar aging candidate words may also be generated based on different aging properties but the same aging search words, or the same or similar aging candidate words are accumulated in the process of continuously updating the aging candidate words; therefore, further, data cleaning operation such as duplicate removal and filtering can be carried out on the generated aging candidate words; the storage pressure of the system on the aging candidate words is reduced, and the recommendation quality of the aging candidate words serving as the candidate words and recommended to the user is improved.

In an embodiment, referring to fig. 5, the method further includes:

s501, determining the life cycle of each aging candidate word corresponding to the sudden aging search word based on a preset time threshold of sudden aging; and/or

S502, determining the life cycle of each aging candidate word corresponding to the periodic aging search word based on the time cycle corresponding to the periodic aging search word;

after the step S103 of constructing the candidate word set of the target user, the method further includes:

s503, determining that the life cycle of any aging candidate word in the candidate word set is finished, and deleting any aging candidate word.

Specifically, the generation process of the aging candidate words may be a continuous activity, and after a period of updating, the existing aging candidate words will be more and more, but the aging candidate words have aging performance, and in order to avoid recommending outdated candidate words to the user and reduce the storage pressure of the system, a life cycle is set for the aging candidate words.

In step S501, processing may be performed based on each aging candidate word corresponding to the bursty aging search word determined in step S201 and/or step S202. Specifically, based on a preset time threshold of the sudden aging, determining a life cycle of each aging candidate word corresponding to the sudden aging search word; the time threshold value can be determined based on the click rate of the search result corresponding to the sudden aging search word in the historical search data; for example, in the click rate of the search result corresponding to the big data analysis paroxysmal aging search word "earthquake", the time based on recall is 6 days, and the time based on accurate search data obtained by statistics is 3 days (the number of days when the click rate reaches the preset click rate threshold); and determining a time threshold (the time threshold is the number of recalling days w1+ the number of accurate days w 2; assumed to be 4 days) based on the weight occupied by the recalling and the counting results; the life cycle of each aging candidate word (such as the latest earthquake message in Shichen county, Ganzui, Sichuan) corresponding to the sudden aging search word "earthquake" is set to be 4 days.

In step S502, processing may be performed based on each aging candidate word corresponding to the periodic aging search word determined in step S203. Specifically, the life cycle of each aging candidate word corresponding to the periodic aging search word is determined based on the time cycle corresponding to the periodic aging search word. The time period is a period corresponding to the periodic aging search term, for example, the periodic aging search term related to the drama, and if the drama is updated once per week, the time period is one week. Optionally, when the life cycle of the aging candidate word corresponding to the periodic aging search word is set, the life cycle may be further set based on the periodic variation time of the click rate in the search result corresponding to the periodic aging search word and the time information corresponding to the time attribute included in the corresponding structured information. For example, the time when the click rate of the search result corresponding to the periodic aging search word "ABC TV play" changes periodically is 2020-04-04, 2020-04-11, and the like; wherein, the time information corresponding to the time attribute included in the corresponding structural information is: the webpage release time is '2020-04-11', and the time-related information in the webpage is 'Saturday update'; the life cycle of each aging candidate word (for example, "ABC series 7 th set") corresponding to the periodic aging search word "ABC series" is set to 7 days. In one embodiment, since the generation time of the aging candidate word with periodic aging is related to the update period, it calculates the start time of the life period as the generation time of the aging candidate word.

In an embodiment, after the step S103 constructs the candidate word set of the target user, the method further includes the step S503 of determining that the life cycle of any aging candidate word in the candidate word set is over, and deleting the aging candidate word. Specifically, such as: in step S501, the aging candidate word "earthquake latest message in shichen county, kataka, kanzi" having a sudden aging is set to have a life cycle of 4 days, and when the life cycle is over, the aging candidate word is aged and deleted in the memory storing the aging candidate word. Such as: in step S502, the aging candidate word "drama XXX set 2" having a periodic aging is set to have a lifetime of 7 days, and when the lifetime ends, the aging candidate word is invalidated and deleted in the memory storing the aging candidate word. Optionally, the method further includes deleting an aging candidate word of which the life cycle is ended in the candidate word set of the target user, so as to avoid recommending an outdated candidate word to the user.

In one embodiment, each aging candidate word corresponding to the long-aging search word has long aging; since the aging candidate words with long timeliness are searched at all time points, no lifecycle or a longer lifecycle (e.g., ten years) may be set for the aging candidate words with long timeliness.

In an embodiment, referring to fig. 6, in step S103, constructing a candidate word set of the target user based on the plurality of aging candidate words and feature information of the target user includes:

s601, calculating the similarity between the plurality of aging candidate words and the characteristic information of the target user respectively, and sequencing the plurality of aging candidate words based on the similarity;

s602, determining at least one aging candidate word in the plurality of aging candidate words to form a candidate word set of the target user based on the sorting result.

Specifically, the feature information of the target user includes information representing a user image acquired after user authorization, such as a user portrait or feature information extracted based on historical search data of the target user; the user portrait abstracts each concrete information of the user into labels, and the labels are used for concreting the user image, including multidimensional information such as age, gender, preference and the like. The similarity is calculated by comparing the similarity of two objects, and is generally determined by calculating the distance between the features of the objects, and if the distance is small, the similarity is large; if the distance is large, the similarity is small. In the embodiment of the present disclosure, the similarity calculation method may use cosine similarity, euclidean distance, "weighted" (minkowski distance), and other methods to calculate the aging candidate words and the feature information of the target user. In step S601, specifically, the similarity between the aging candidate words and the feature information of the target user is calculated, and after the similarity value corresponding to each aging candidate word is obtained, the aging candidate words are ranked based on the similarity value. In one embodiment, the method can be realized through a ranking stage of a recommendation model, such as CTR estimation, and the click probability of each aging candidate word by a target user is predicted in massive aging candidate words so as to perform ranking; the recommendation model comprises two stages (a recall stage and a sorting stage), wherein in the recall stage, a small candidate set (hundreds to thousands of search terms) is selected from the tens of millions of search terms according to historical search data, for example, the search terms of the whole network are filtered and checked layer by layer to obtain the search terms; in the sorting stage, more accurate calculation is performed on the basis of the recall stage, each search term can be accurately scored, and then the search term (a dozen of search terms) most interested by the target user is selected from thousands of search terms. In the embodiment of the disclosure, the aging candidate words are generated based on the aging search words and the structural information corresponding to the aging search words to replace the recall stage in the recommendation model, and in the sorting stage, each aging candidate word is accurately scored by combining the characteristic information of the target user on the basis of the aging candidate words generated in step S102, so that a small number of aging candidate words with the largest click rate of the target user are screened out through step S602 and combined into the candidate word set of the target user.

In step S602, determining that at least one aging candidate word is combined into a candidate word set of the target user from the aging candidate words based on the sorting result; in one embodiment, the sorting is generally in descending order; and combining the aging candidate words with high similarity into a candidate word set of the target user based on the result after descending sorting. When screening the aging candidate words based on the sorting result, a preset value can be set, for example, the aging candidate words with the similarity of more than 80% are determined as candidate words forming the target user, or the aging candidate words with the top 20% in descending sorting are determined as candidate words of the target user; can be adjusted according to actual conditions.

In an embodiment, the step S601 of calculating the similarity between each of the plurality of aging candidate words and the feature information of the target user includes: extracting preset features from feature information of a target user; and respectively calculating the similarity of the plurality of aging candidate words and preset characteristics, and determining the calculation result as the similarity of the plurality of aging candidate words and the characteristic information of the target user.

Specifically, preset features such as feature engineering in a recommendation model are extracted from feature information of a target user, and the preset features are set based on similarity of a calculated aging candidate word and the feature information, such as common discrete features including user preferences. When the similarity between the plurality of aging candidate words and the preset characteristics is calculated respectively, a Logistic Regression (LR) model can be adopted to predict the probability of clicking each aging candidate word by the target user. Specifically, the similarity may also be calculated by using cosine similarity, euclidean distance, "weighted" (minkowski distance), or other methods; the method for calculating the similarity between the aging candidate words and the preset features by adopting a cosine similarity calculation method comprises the following steps: calculating the cosine values of the word vectors of the aging candidate words and the feature vectors of the preset features; when the word vector of the aging candidate word is calculated, the word vector can be determined by adopting a word segmentation method of character string matching or statistics. Calculating the similarity between the aging candidate words and the preset characteristics by adopting a calculation method of Euclidean distance, wherein the calculation method comprises the following steps: marking data points of each characteristic on a coordinate marked with the preset characteristic, and calculating the Euclidean distance from the data point corresponding to the preset characteristic to the aging candidate word by using the coordinate; and when the preset feature comprises a plurality of data points, marking the data points forming the multi-dimensional vector.

In an embodiment, referring to fig. 7, in step S103, constructing a candidate word set of the target user based on the aging candidate words and feature information of the target user includes:

s701, integrating the plurality of aging candidate words with preset recommendation candidate words;

s702, calculating the similarity between the integrated candidate words and the characteristic information of the target user, and sequencing the candidate words based on the similarity;

s703, determining at least one candidate word in the integrated candidate words to form a candidate word set of the target user based on the sorting result.

Specifically, the preset recommended candidate word is an existing search word, and in the recommendation model, the preset recommended candidate word can be understood as a plurality of existing search words in a model recall stage, which can form a recall set; when the aging candidate words generated in step S102 are integrated with the preset recommended candidate words, the integrated candidate words include a plurality of aging candidate words and a plurality of recommended candidate words, and it can also be understood that the plurality of aging candidate words are integrated into the recall set. In the sorting stage, the candidate words in the recall set are specifically screened and sorted again, for example, the probability that a candidate word is selected by a target user as a search word for searching is determined, the higher the probability selected by the user is, the earlier the sorting is, and finally the candidate words ranked earlier are determined to be combined into the candidate word set of the target user. Steps S701 to S703 correspond to step S601, and the difference is that the recall set of the recommendation model is inherited in S701, that is, the generated aging candidate words are merged into the recall set in combination with the recall stage of the recommendation model, and participate in the ranking stage of the recommendation model.

Specifically, the sorting phase comprises: extracting preset features from feature information of a target user; calculating the similarity between each candidate word in the integrated candidate words and a preset feature, sorting the candidate words based on the similarity, and determining at least one candidate word in the integrated candidate words to form a candidate word set of the target user based on a sorting result; the integrated candidate words comprise a plurality of aging candidate words and recommendation candidate words.

In one embodiment, when the aging candidate words are continuously updated, a plurality of aging candidate words are continuously merged into the recall set. In the sorting stage, the top candidate words may include an aging candidate word and a recommendation candidate word, or may include only an aging candidate word or only a recommendation candidate word. Specifically, a plurality of search terms are stored in the database, and the search terms are existing and are irrelevant to the individuation of the user; the method and the device for recalling the search words adopt the recommendation model to recall the search words, namely, a plurality of recommendation candidate words related to historical search behaviors of a plurality of users are screened out, and a recall set is formed by the recommendation candidate words. The recall set may also be continuously updated, with the rate of update depending on the recall efficiency of the user using the search data and recommendation model of the search engine. The sequencing stage of the recommendation model may adopt a DNN model, an FM model, and the like, and those skilled in the art can determine the sequencing stage according to the actual situation, which is not described herein.

In an embodiment, referring to fig. 8, after the step S103 constructs the candidate word set of the target user, the method further includes:

s801, pushing the candidate words in the candidate word set of the target user to the client corresponding to the target user.

Specifically, the target user may be any user using a search engine, and after the candidate word set of the target user is constructed and the target user initiates a search request, the candidate words in the candidate word set may be pushed to a client corresponding to the target user, where all the candidate words in the candidate word set may be pushed, or a preset number of candidate words may be extracted from the candidate word set and pushed to the client corresponding to the target user based on a preset number of the pushed candidate words (for example, the number of the pushed candidate words may be displayed by the client).

In an embodiment, referring to fig. 9, step S801 sends candidate words in the candidate word set of the target user to a client corresponding to the target user, where the step includes any one of the following steps:

s901, pushing candidate words related to the bursty aging in the candidate word set of the target user to a client corresponding to the target user within a preset time interval at a first preset pushing frequency;

s902, the candidate words related to the periodic aging in the candidate word set of the target user are pushed to the client corresponding to the target user at a second preset pushing frequency within a time period corresponding to the periodic aging.

Specifically, the candidate words with various timeliness are included in the candidate word set constructed in step S103, such as candidate words with bursty timeliness, with periodic timeliness, with long timeliness, and the like.

In step S901, processing is performed on candidate words related to bursty aging in the candidate word set. Specifically, when a target user initiates a search request and the initiation time is within a predetermined time interval, the relevant candidate words are pushed to a client corresponding to the target user at a first predetermined pushing frequency. The preset time interval can be set to be one day, several hours or several seconds, and related candidate words are continuously pushed to the target user.

In step S902, a process is performed on a candidate word related to the periodic aging in the candidate word set. Specifically, when the target user initiates a search request and the initiation time is within a time period corresponding to the periodic aging, the relevant candidate words are pushed to the client corresponding to the target user at the second predetermined pushing frequency. If the candidate word "high-level entrance for higher-level entrance for lower-level entrance for higher.

Referring to fig. 10, a schematic structural diagram of a candidate word constructing apparatus 100 for search according to an embodiment of the present disclosure is provided, where the candidate word constructing apparatus 100 for search according to the embodiment of the present disclosure may include:

a determining module 1001 configured to determine a plurality of aged search terms associated with time based on historical search data;

the generating module 1002 is configured to obtain structural information corresponding to the aging search word, and generate a plurality of aging candidate words based on the aging search word and the structural information corresponding to the aging search word; the structured information comprises a plurality of entity information formed by extracting information from webpage data;

a constructing module 1003, configured to construct a candidate word set of the target user based on the multiple aging candidate words and feature information of the target user.

In an embodiment, the determining module 1001 includes at least one of the following:

the device comprises a first extraction unit, a second extraction unit and a third extraction unit, wherein the first extraction unit is used for acquiring multiple times of search behavior data in a preset time interval from historical search data, and extracting a keyword from the search behavior data corresponding to a historical search word as a sudden aging search word if the number of the search behavior data aiming at the same historical search word in the preset time interval exceeds a first preset number threshold;

a second extraction unit, configured to obtain multiple search behavior data in a preset time interval from the historical search data, and if it is identified that any historical search word meets a predetermined bursty aging characteristic and the number of search behavior data for any historical search word in the preset time interval exceeds a second preset number threshold, extract a keyword from the search behavior data corresponding to any historical search word as a bursty aging search word;

and the third extraction unit is used for extracting keywords from the search behavior data corresponding to the historical search words as periodic aging search words if the periodic search behavior data for the same historical search word are extracted from the historical search data and the number of the search behavior data in each period exceeds a third preset number threshold.

In an embodiment, the generating module 1002 includes at least one of the following:

the first acquisition unit is used for inquiring from a database storing a plurality of pieces of structural information according to the periodic aging search word and determining the structural information containing the periodic aging search word as the structural information corresponding to the aging search word;

the second acquisition unit is used for inquiring from a database storing a plurality of structural information according to the sudden aging search term and determining the structural information containing the sudden aging search term as the structural information corresponding to the aging search term;

a third obtaining unit, configured to query from a database storing a plurality of pieces of structural information according to the bursty aging search term and trending query topic information associated with the bursty aging search term, and determine partial structural information related to the trending query topic information in the structural information including the bursty aging search term as structural information corresponding to the aging search term;

the first generation unit is used for generating a plurality of aging candidate words based on the sudden aging search words, each entity keyword included in the structural information corresponding to the sudden aging search words and a preset lexical item of sudden aging;

and the second generating unit is used for generating a plurality of aging candidate words based on the periodic aging search words and each entity keyword included in the structural information corresponding to the periodic aging search words.

In one embodiment, the apparatus 100 further comprises:

the first setting module is used for determining the life cycle of each aging candidate word corresponding to the sudden aging search word based on a preset time threshold of sudden aging; and/or

The second setting module is used for determining the life cycle of each aging candidate word corresponding to the periodic aging search word based on the time cycle corresponding to the periodic aging search word;

after the constructing the candidate word set of the target user, the apparatus 100 further includes:

and the deleting module is used for determining that the life cycle of any aging candidate word in the candidate word set is finished and deleting the aging candidate word.

In an embodiment, the building module 1003 includes:

the first calculating unit is used for calculating the similarity between the plurality of aging candidate words and the characteristic information of the target user respectively and sequencing the plurality of aging candidate words based on the similarity;

and the first sequencing unit is used for determining at least one aging candidate word from the plurality of aging candidate words to form the candidate word set of the target user based on the sequencing result.

In an embodiment, the building module 1003 includes:

the integration unit is used for integrating the plurality of aging candidate words with preset recommendation candidate words;

the second calculation unit is used for calculating the similarity between the integrated candidate words and the feature information of the target user and sequencing the candidate words based on the similarity;

and the second sorting unit is used for determining at least one candidate word from the integrated candidate words to form the candidate word set of the target user based on a sorting result.

In one embodiment, the apparatus 100 further comprises:

the pushing module is used for pushing the candidate words in the candidate word set of the target user to the client corresponding to the target user;

the pushing module comprises any one of the following units:

the first pushing unit is used for pushing candidate words related to the bursty timeliness in the candidate word set of the target user to a client corresponding to the target user within a preset time interval at a first preset pushing frequency;

and the second pushing unit is used for pushing the candidate words related to the periodic aging in the candidate word set of the target user to the client corresponding to the target user at a second preset pushing frequency in a time period corresponding to the periodic aging.

The candidate word construction device for searching according to the embodiment of the present disclosure may execute a candidate word construction method for searching provided by the embodiment of the present disclosure, and the implementation principles thereof are similar, the actions performed by each module in the candidate word construction device for searching according to the embodiments of the present disclosure correspond to the steps in the candidate word construction method for searching according to the embodiments of the present disclosure, and for the detailed functional description of each module of the candidate word construction device for searching, reference may be specifically made to the description in the corresponding candidate word construction method for searching shown in the foregoing, and details are not repeated here.

Referring now to FIG. 11, a block diagram of an electronic device (e.g., server) 600 suitable for use in implementing embodiments of the present disclosure is shown. The server in the embodiments of the present disclosure may include, but is not limited to, a device such as a computer and the like. The electronic device shown in fig. 11 is only an example, and should not bring any limitation to the functions and the scope of use of the embodiments of the present disclosure.

The electronic device includes: a memory and a processor, wherein the processor may be referred to as the processing device 601 hereinafter, and the memory may include at least one of a Read Only Memory (ROM)602, a Random Access Memory (RAM)603 and a storage device 608 hereinafter, which are specifically shown as follows:

as shown in fig. 11, electronic device 600 may include a processing means (e.g., central processing unit, graphics processor, etc.) 601 that may perform various appropriate actions and processes in accordance with a program stored in a Read Only Memory (ROM)602 or a program loaded from a storage means 608 into a Random Access Memory (RAM) 603. In the RAM 603, various programs and data necessary for the operation of the electronic apparatus 600 are also stored. The processing device 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input/output (I/O) interface 605 is also connected to bus 604.

Generally, the following devices may be connected to the I/O interface 605: input devices 606 including, for example, a touch screen, touch pad, keyboard, mouse, camera, microphone, accelerometer, gyroscope, etc.; output devices 607 including, for example, a Liquid Crystal Display (LCD), a speaker, a vibrator, and the like; storage 608 including, for example, tape, hard disk, etc.; and a communication device 609. The communication means 609 may allow the electronic device 600 to communicate with other devices wirelessly or by wire to exchange data. While fig. 11 illustrates an electronic device 600 having various means, it is to be understood that not all illustrated means are required to be implemented or provided. More or fewer devices may alternatively be implemented or provided.

In particular, according to an embodiment of the present disclosure, the processes described above with reference to the flowcharts may be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program product comprising a computer program carried on a non-transitory computer readable medium, the computer program containing program code for performing the method illustrated by the flow chart. In such an embodiment, the computer program may be downloaded and installed from a network via the communication means 609, or may be installed from the storage means 608, or may be installed from the ROM 602. The computer program, when executed by the processing device 601, performs the above-described functions defined in the methods of the embodiments of the present disclosure.

It should be noted that the computer readable medium in the present disclosure can be a computer readable signal medium or a computer readable medium or any combination of the two. A computer readable medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples of the computer readable medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the present disclosure, a computer readable medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. In contrast, in the present disclosure, a computer readable signal medium may comprise a propagated data signal with computer readable program code embodied therein, either in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may be any computer readable medium that is not a computer readable medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to: electrical wires, optical cables, RF (radio frequency), etc., or any suitable combination of the foregoing.

In some embodiments, the clients, servers may communicate using any currently known or future developed network protocol, such as HTTP (HyperText transfer protocol), and may be interconnected with any form or medium of digital data communication (e.g., a communications network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), the Internet (e.g., the Internet), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future developed network.

The computer readable medium may be embodied in the electronic device; or may exist separately without being assembled into the electronic device.

The computer readable medium carries one or more programs which, when executed by the electronic device, cause the electronic device to perform the steps of: determining a plurality of aged search terms associated with time based on historical search data; acquiring structural information corresponding to the aging search word, and generating a plurality of aging candidate words based on the aging search word and the structural information corresponding to the aging search word; the structured information comprises a plurality of entity information formed by extracting information from webpage data; and constructing a candidate word set of the target user based on the plurality of aging candidate words and the characteristic information of the target user.

Computer program code for carrying out operations for the present disclosure may be written in any combination of one or more programming languages, including but not limited to an object oriented programming language such as Java, Smalltalk, C + +, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet service provider).

The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

The modules or units described in the embodiments of the present disclosure may be implemented by software or hardware. Where the name of a module or unit does not in some cases constitute a limitation on the unit itself, for example, the determination module may also be described as a "module for determining a plurality of aged search terms associated with time based on historical search data".

The functions described herein above may be performed, at least in part, by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field Programmable Gate Arrays (FPGAs), Application Specific Integrated Circuits (ASICs), Application Specific Standard Products (ASSPs), systems on a chip (SOCs), Complex Programmable Logic Devices (CPLDs), and the like.

In the context of this disclosure, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

According to one or more embodiments of the present disclosure, there is provided a candidate word construction method for search, including: determining a plurality of aged search terms associated with time based on historical search data; acquiring structural information corresponding to the aging search word, and generating a plurality of aging candidate words based on the aging search word and the structural information corresponding to the aging search word; the structured information comprises a plurality of entity information formed by extracting information from webpage data; and constructing a candidate word set of the target user based on the plurality of aging candidate words and the characteristic information of the target user.

In one embodiment, the determining a plurality of aged search terms associated with a time based on historical search data includes at least one of:

acquiring multiple times of search behavior data in a preset time interval from historical search data, and if the number of the search behavior data aiming at the same historical search word in the preset time interval exceeds a first preset number threshold, extracting a keyword from the search behavior data corresponding to the historical search word to be used as a sudden aging search word;

acquiring multiple times of search behavior data in a preset time interval from historical search data, and if any historical search word is identified to be in accordance with preset paroxysmal timeliness characteristics and the number of the search behavior data aiming at any historical search word in the preset time interval exceeds a second preset number threshold, extracting a keyword from the search behavior data corresponding to any historical search word to be used as a paroxysmal timeliness search word;

if periodic search behavior data aiming at the same historical search word are extracted from the historical search data, and the number of the search behavior data in each period exceeds a third preset number threshold, extracting keywords from the search behavior data corresponding to the historical search word as periodic aging search words.

In an embodiment, the obtaining of the structured information corresponding to the aging search term includes at least one of:

according to the periodic aging search word, querying a database in which a plurality of structural information are stored, and determining the structural information containing the periodic aging search word as the structural information corresponding to the aging search word;

according to the sudden aging search word, inquiring from a database in which a plurality of structural information are stored, and determining the structural information containing the sudden aging search word as the structural information corresponding to the aging search word;

inquiring from a database storing a plurality of structural information according to the sudden aging search words and the hot query topic information associated with the sudden aging search words, and determining part of the structural information which contains the structural information of the sudden aging search words and has a relationship with the hot query topic information as the structural information corresponding to the aging search words;

In an embodiment, the generating a plurality of aging candidate words based on the aging search word and the structural information corresponding to the aging search word includes at least one of:

generating a plurality of aging candidate words based on the sudden aging search words, each entity keyword included in the structural information corresponding to the sudden aging search words and a preset lexical item of sudden aging;

and generating a plurality of aging candidate words based on the periodic aging search words and each entity keyword included in the structural information corresponding to the periodic aging search words.

In an embodiment, the method further comprises:

determining the life cycle of each aging candidate word corresponding to the sudden aging search word based on a preset time threshold of sudden aging; and/or

Determining the life cycle of each aging candidate word corresponding to the periodic aging search word based on the time cycle corresponding to the periodic aging search word;

after the candidate word set of the target user is constructed, the method further comprises the following steps:

and determining that the life cycle of any aging candidate word in the candidate word set is finished, and deleting any aging candidate word.

In an embodiment, the constructing the candidate word set of the target user based on the plurality of aging candidate words and the feature information of the target user includes:

calculating the similarity between the plurality of aging candidate words and the characteristic information of the target user respectively, and sequencing the plurality of aging candidate words based on the similarity;

determining at least one aging candidate word in the plurality of aging candidate words to form a candidate word set of the target user based on the sorting result.

integrating the plurality of aging candidate words with preset recommendation candidate words;

calculating the similarity of the integrated candidate words and the characteristic information of the target user, and sequencing the candidate words based on the similarity;

and determining at least one candidate word from the integrated candidate words to form a candidate word set of the target user based on the sorting result.

In an embodiment, after the constructing the candidate word set of the target user, the method further includes:

pushing the candidate words in the candidate word set of the target user to a client corresponding to the target user;

the pushing of the candidate words in the candidate word set of the target user to the client corresponding to the target user includes any one of the following:

pushing candidate words related to the paroxysmal timeliness in the candidate word set of the target user to a client corresponding to the target user within a preset time interval at a first preset pushing frequency;

and pushing the candidate words related to the periodic aging in the candidate word set of the target user to the client corresponding to the target user at a second preset pushing frequency in a time period corresponding to the periodic aging.

According to one or more embodiments of the present disclosure, there is provided a candidate word construction apparatus for search, including:

a determination module to determine a plurality of aged search terms associated with time based on historical search data;

the generating module is used for acquiring structural information corresponding to the aging search words and generating a plurality of aging candidate words based on the aging search words and the structural information corresponding to the aging search words; the structured information comprises a plurality of entity information formed by extracting information from webpage data;

and the constructing module is used for constructing a candidate word set of the target user based on the plurality of aging candidate words and the characteristic information of the target user.

In an embodiment, the determining module comprises at least one of:

In one embodiment, the generation module includes at least one of the following:

In one embodiment, the apparatus further comprises:

In one embodiment, the building block includes:

In one embodiment, the apparatus further comprises:

the pushing module comprises any one of the following units:

The foregoing description is only exemplary of the preferred embodiments of the disclosure and is illustrative of the principles of the technology employed. It will be appreciated by those skilled in the art that the scope of the disclosure herein is not limited to the particular combination of features described above, but also encompasses other embodiments in which any combination of the features described above or their equivalents does not depart from the spirit of the disclosure. For example, the above features and (but not limited to) the features disclosed in this disclosure having similar functions are replaced with each other to form the technical solution.

Further, while operations are depicted in a particular order, this should not be understood as requiring that such operations be performed in the particular order shown or in sequential order. Under certain circumstances, multitasking and parallel processing may be advantageous. Likewise, while several specific implementation details are included in the above discussion, these should not be construed as limitations on the scope of the disclosure. Certain features that are described in the context of separate embodiments can also be implemented in combination in a single embodiment. Conversely, various features that are described in the context of a single embodiment can also be implemented in multiple embodiments separately or in any suitable subcombination. Although the subject matter has been described in language specific to structural features and/or methodological acts, it is to be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are disclosed as example forms of implementing the claims.

Claims

1. A candidate word construction method for searching is characterized by comprising the following steps:

determining a plurality of aged search terms associated with time based on historical search data;

acquiring structural information corresponding to the aging search word, and generating a plurality of aging candidate words based on the aging search word and the structural information corresponding to the aging search word; the structured information comprises a plurality of entity information formed by extracting information from webpage data;

and constructing a candidate word set of the target user based on the plurality of aging candidate words and the characteristic information of the target user.

2. The method of claim 1, wherein determining a plurality of aged search terms associated with time based on historical search data comprises at least one of:

3. The method of claim 2, wherein obtaining the structured information corresponding to the aged search term comprises at least one of:

4. The method of claim 2, wherein generating a plurality of aging candidate words based on the aging search word and the structured information corresponding to the aging search word comprises at least one of:

5. The method of claim 2, further comprising:

6. The method of claim 1, wherein the constructing the target user's candidate word set based on the plurality of aging candidate words and the target user's feature information comprises:

7. The method of claim 1, wherein the constructing the target user's candidate word set based on the plurality of aging candidate words and the target user's feature information comprises:

8. The method of claim 1, wherein after constructing the set of candidate words for the target user, the method further comprises:

9. A candidate word construction apparatus for search, comprising:

10. An electronic device, comprising:

the electronic device comprises a memory and a processor;

the memory has stored therein a computer program;

the processor, when executing the computer program, is configured to perform the method of any of claims 1-8.

11. A computer-readable medium, on which a computer program is stored, which, when being executed by a processor, carries out the method of any one of claims 1 to 8.