CN115269959A - Search association recommendation method and device, electronic equipment and storage medium - Google Patents
Search association recommendation method and device, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN115269959A CN115269959A CN202210870530.6A CN202210870530A CN115269959A CN 115269959 A CN115269959 A CN 115269959A CN 202210870530 A CN202210870530 A CN 202210870530A CN 115269959 A CN115269959 A CN 115269959A
- Authority
- CN
- China
- Prior art keywords
- search
- word
- relevant
- scene
- words
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9532—Query formulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/274—Converting codes to words; Guess-ahead of partial word inputs
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The application relates to the technical field of computer application, in particular to a search association recommendation method, a search association recommendation device, electronic equipment and a storage medium, wherein the method comprises the following steps: generating a relevant recommended word bank, wherein the word bank comprises a relevant word set and scene relevancy information of different relevant words; acquiring a search request input by a user, wherein the search request carries a search keyword; performing intention identification processing on the search request to acquire a search intention; generating a relevant word recommendation set associated with the search keyword and the search intention by using the relevant recommendation word library; and displaying a relevant word recommendation set with relevant scene information of relevant words. According to the method and the device, the associated recommended words recommended to the user can be matched with the user requirements more accurately.
Description
Technical Field
The present application relates to the field of computer application technologies, and in particular, to a search association recommendation method and apparatus, an electronic device, and a storage medium.
Background
At present, in the operation of searching for comprehensive information in a search page or a search bar by a user, a system generally associates some data results according to a search keyword of the user to help the user directly select an associated recommended word under the condition of inputting a short keyword, so that the input steps of the user are reduced, the search input time of the user is shortened, and the user is helped to quickly find the content to be searched.
In the related technology, after preprocessing work such as format check, format conversion, error correction, word segmentation and the like is carried out on the search keywords, related associative recommended words are obtained and displayed to a user.
In practice, the inventors found that there are at least the following problems in this technique:
the recommended associated recommended words are not distinguished according to scenes, so the associated recommended words recommended to the user are sometimes confused, and the recommended words recommended to the user cannot be more accurately matched with the user requirements.
Disclosure of Invention
In order to solve the above problems, the present application provides a search association recommendation method, device, electronic device, and storage medium, so that an associated recommended word recommended to a user more accurately matches a user requirement.
In a first aspect, the present application provides a search association recommendation method, which adopts the following technical solution:
a search association recommendation method, the method comprising:
generating a relevant recommended word bank, wherein the word bank comprises a relevant word set and scene relevancy information of different relevant words;
acquiring a search request input by a user, wherein the search request carries a search keyword;
performing intention identification processing on the search request to acquire a search intention;
generating a relevant word recommendation set associated with the search keyword and the search intention by using the relevant recommendation word bank;
and displaying the relevant word recommendation set with relevant scene information of the relevant words.
By the technical scheme, the relevant words matched with the search keywords and the search intention are obtained from the relevant recommended word bank, and the relevant recommended word set is obtained and displayed, and the scene information with the relevant words is displayed.
In some embodiments, the generating the associated recommended thesaurus specifically includes:
establishing a related recommended word bank, wherein the related recommended word bank comprises a related word set formed by related words;
and determining scene matching distribution of each relevant word in each preset scene according to the correlation degree of the relevant word and different preset scenes.
By the technical scheme, the associated recommended word library is established, the scene matching scores corresponding to each to-be-selected associated word in the associated recommended word library under each preset scene are obtained through calculation, and whether each to-be-selected associated word can be matched under a certain scene or not can be determined through the values of the scene matching scores. The data in the associated recommended thesaurus can be dynamically updated.
In some embodiments, the determining a scene matching score of each relevant word in each preset scene further includes any of the following manners or any combination thereof:
a) Performing data cleaning and data enhancement processing on the associated recommended word bank to generate different scene matching scores of different associated words;
b) Generating scene matching scores for different associated words in the associated recommended word library based on the historical behaviors of the user and the associated information of the associated words;
c) And generating different scene matching scores of different associated words in the associated recommended word library through an expert system.
By the technical scheme, the scene matching score can be determined by one or more methods.
In some embodiments, further comprising: and determining the scene of each relevant word according to the scene matching distribution of different relevant words.
By the technical scheme, the scene matching threshold can be set according to requirements, the scene matching score of the scene corresponding to the associated word is larger than the preset scene matching threshold, and the scene can be used as the scene to which the associated word belongs.
In some embodiments, the generating, by using the associated recommendation thesaurus, an associated word recommendation set associated with the search keyword and the search intention specifically includes:
determining a preset scene associated with the search intention;
acquiring a relevant word to be selected matched with the associated preset scene according to the scene relevancy information of the relevant word;
determining the associated matching points of the associated word to be selected and the search keyword;
and generating a relevant word recommendation set according to the relevant matching scores of the relevant words to be selected, wherein the relevant word recommendation set comprises relevant scene information of the relevant words.
Through the technical scheme, the related scene is determined through the search intention, and the related words to be selected in the related scene are matched with the search related words, so that the related word recommendation set with the information of the scene to which the related words belong is generated. The type and the recommendation number of the displayed associated words can be adjusted through setting a preset scene.
In some embodiments, the generating the set of related-word recommendations further comprises:
acquiring the search heat degree and/or content comprehensive score of the associated word to be selected;
generating a comprehensive score of the associated word to be selected according to the search heat degree and/or the content comprehensive score of the associated word to be selected and the associated matching score of the associated word to be selected and the search keyword;
and sequencing the associated words to be selected according to the comprehensive scores of the associated words to be selected.
According to the technical scheme, the generated related word recommendation set is rearranged through the search heat information and/or the content comprehensive information of the related words, so that the rearranged related word recommendation set is obtained.
In some embodiments, the generating of the relevant word recommendation set associated with the search keyword and the search intention using the relevant recommendation thesaurus further comprises:
and generating a relevant word recommendation set according to the historical search words of the user and the search heat information of the historical search words.
By the technical scheme, the historical search words input by the user can be used as candidate words recommended by association, and the requirements of the user can be met.
In a second aspect, the present application provides a search association recommendation apparatus, which adopts the following technical solutions:
a search association recommendation apparatus comprising:
the word bank generating module is used for generating a relevant recommended word bank, and the word bank comprises a relevant word set and scene relevancy information of different relevant words;
the system comprises a search request acquisition module, a search query processing module and a search query processing module, wherein the search request acquisition module is used for acquiring a search request input by a user, and the search request carries a search keyword;
the intention identification module is used for carrying out intention identification processing on the search request and acquiring a search intention;
the recommendation generation module is used for generating a relevant word recommendation set associated with the search keyword and the search intention by using the relevant recommendation word bank;
and the recommendation display module is used for displaying the relevant word recommendation set with relevant scene information of relevant words.
In a third aspect, the present application provides an electronic device, which adopts the following technical solutions:
an electronic device, comprising:
at least one processor;
storage means for storing at least one computer program;
when executed by the at least one processor, the at least one computer program causes the at least one processor to implement the method of the above-described aspects.
In a fourth aspect, the present application provides a computer-readable storage medium, which adopts the following technical solutions:
a computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the method of the above-mentioned solution.
In summary, the present application includes at least one of the following beneficial technical effects:
1. the associated recommended words recommended to the user can distinguish a plurality of scenes and are displayed according to the scenes respectively, and requirements of the user can be matched more accurately.
2. On the basis of recommending the associated recommended words according to the preset scene type, historical search scenes can be added, namely, the dynamically-changed historical search associated recommended words are provided according to the historical input of the user, and the user experience is improved.
3. The word stock can be dynamically updated according to requirements, the required scene type is set, and the recommendation type and the number of the associated recommended words are flexibly adjusted.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the description below are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings without creative efforts.
Fig. 1 is a flowchart illustrating a method for generating an associated recommended word library according to an embodiment of the present application;
fig. 2 is a schematic overall flowchart of a search association recommendation method according to an embodiment of the present application;
fig. 3 is a flowchart illustrating a method for generating a recommendation set of related words according to an embodiment of the present application;
fig. 4 is a schematic diagram of a framework of a search association recommendation apparatus according to an embodiment of the present application.
Detailed Description
In order to make the objects, features and advantages of the present application more obvious and understandable, the technical solutions in one embodiment of the present application will be clearly and completely described below with reference to the accompanying drawings in one embodiment of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
The present application is described in further detail below with reference to figures 1 to 4.
The method for generating the associated recommended word stock as shown in fig. 1 may implement the following steps:
101. and establishing an associated recommended word bank, wherein the associated recommended word bank comprises an associated word set formed by associated words.
The meaning of the related words is as follows: containing search keywords or words having similar meanings to the search keywords. For example, searching for the word "appliance", including the word "appliance", or words having a meaning close to that of "appliance", includes "XX appliances limited", "XX appliances manufacturing limited", "XX appliances limited", "an appliance box", "an appliance apparatus", "a charger", "appliance apparatus", "charger", "relay"; these are all relevant words of "electric appliances" in the relevant recommended word library.
102. And determining scene matching distribution of each associated word in each preset scene according to the correlation degree of the associated word and different preset scenes.
Each related word has a different degree of association with different preset scenes, for example, in the above example, the degree of association between an electric company and an enterprise scene is the highest, and the degree of association between words such as electric appliances and chargers and product and patent scenes is higher. And each associated word has a scene matching score in each preset scene.
In the embodiment of the application, the determining of the scene matching score of each relevant word in each preset scene may further implement any of the following manners or any combination manner thereof:
a) Performing data cleaning and data enhancement processing on the associated recommended word bank to generate different scene matching scores of different associated words;
data processing and expansion can be performed on the existing word stock, and the assigned scene matching score is higher than a preset scene matching threshold value.
b) Generating scene matching scores for different associated words in the associated recommended word library based on the historical behaviors of the user and the associated information of the associated words;
based on the historical behaviors of the user (clicking and other operations), scene matching assignment is carried out on the scene corresponding to the associated word associated with the historical behaviors of the user, and the scene matching assignment of the assignment is higher than a preset scene matching threshold.
c) And generating different scene matching scores of different associated words in the associated recommended word bank through an expert system.
And (3) carrying out scene matching score assignment on the scene type corresponding to the associated word through an expert system (expert scoring), wherein the assigned scene matching score is higher than a preset scene matching threshold, and the method can be used for processing the vocabulary in the specific field.
And (c) executing any one or any combination of the three modes a, b and c, after the associated recommended word base is established, continuously and dynamically adjusting the scene matching score of the associated word in the mode b, wherein the dynamically adjusted score can be adjusted each time according to a preset assignment rule, for example, the dynamically adjusted score can be increased by 5-10 scores each time effective clicking is performed.
When the scene matching time is assigned to the scene irrelevant to the relevant words in the relevant recommended word bank, a lower score (lower than a preset scene matching threshold value) can be assigned according to a preset assignment rule, and if the scene matching time is associated in the subsequent user historical behavior process, the scene matching time can be dynamically adjusted.
In the embodiment of the application, the scene of each relevant word is determined according to the scene matching distribution of different relevant words.
The scene matching score may determine the scene matching threshold value in accordance with different situations, and for example, a corresponding scene in which the scene matching score of each related word is 70 or 80 or more may be set as the scene to which the related word belongs. The related word belongs to one or more scenes.
For example, the term "artificial intelligence" can be used in this application in a plurality of scene types: enterprise class, product class, and patent class. Such as "deep learning," the context type to which this term is applied may be patent only.
And when the scene matching score exceeds a set threshold, the scene type can be considered as the scene type of the associated word to be selected. The data in the associated recommended word bank is updated at regular time, and the data is relatively static when the data is not updated.
In one embodiment of the application, the associated recommended word library is constructed through hive and elastic search, data T +1 updating is performed through hive library integrated data at regular time tasks every day, the associated recommended word library is updated, modified and deleted through a data warehouse platform under big data, a word library operation platform of a data service middle platform and kafka data real-time stream, hour-level updating is supported, and dirty data can be cleaned and filtered based on content portraits and filtering rules. The word bank warehousing rule is as follows: 1) No special character string is contained; 2) The user searches for click behaviors on the first screen; 3) The number of searches used on a single day is greater than n; 4) When the short word and the long word search click results are the same, the long word is taken; 5) And maintaining the word bank manually and periodically. And each service data source integrates a data hive library, and regularly schedules and updates the associated recommended word library every day.
The search association recommendation method shown in fig. 2 may implement the following steps:
201. generating a related recommended word bank, wherein the word bank comprises a related word set and scene relevancy information of different related words;
202. the method comprises the steps of obtaining a search request input by a user, wherein the search request carries search keywords.
The method comprises the steps of obtaining a search request of a user, wherein the search request can be in various forms, such as characters, voice, graphs and the like, and the characters can be pinyin, chinese characters, english, pure letters and the like.
The search request comprises search keywords which can be directly used without pretreatment after format check; preprocessing may also be required to obtain the adjusted search keywords. The method comprises the steps of preprocessing search words input by a user, wherein the preprocessing comprises sensitive word processing, abnormal symbol escape, search keyword extraction, search keyword error correction processing and the like, and correcting irregular input and error input.
The search keyword error correction process may employ a chinese error correction model, such as a pycorrector model.
The Chinese error correction comprises the first step of error detection and the second step of error correction.
1) The error detection part cuts words through a Chinese word segmentation device, and because sentences contain wrongly-written characters, word cutting results are always wrong in segmentation, so that errors are detected from two aspects of character granularity and word granularity, suspected error results of the two granularities are integrated, and a suspected error position candidate set is formed;
2) And the error correction part is used for traversing all suspected error positions, replacing words in the error positions by using similar dictionaries and similar dictionaries, calculating sentence confusion degree through a language model, and comparing and sequencing results of all candidate sets to obtain the optimal corrected words.
203. And performing intention identification processing on the search request to acquire a search intention.
And judging the search intention of the user through an intention recognition model. The intention recognition model employs a commonly used text classification model. The search intent is the reason or purpose for which the user performs the search. For example, a user inputs "appliance" and the search is intended to obtain business related information (appliance related businesses), product related information (appliance related products), and patent related information (appliance related patents). Through the identification of the search intention of the user, the recommended associated recommended words can be more fit with the behaviors of the user.
1) Acquiring data, manually marking a small part of data, and crawling most of training data on a network;
2) The method comprises the steps of carrying out pre-training processing such as special character processing, missing value processing and word conversion into a vector (word 2 vec) on data, solving the average value of the processed vector, solving the problem of out of vocabulary characters, and carrying out low-dimensional vector (embedding) feature addition based on semantic relation and position relation after processing;
3) And connecting a random inactivation (dropout) layer to the obtained processed low-dimensional vector layer based on a convolutional neural network (cnn) training model parameter, and calculating a scene to which the search intention belongs by using a cross entropy loss function for a final output layer.
And when the number of the identified search intents is multiple, sorting the search intents according to the association degree of the search intents and the search keywords.
204. And generating a relevant word recommendation set associated with the search keyword and the search intention by using the relevant recommendation word library.
The step can judge which scene type or types the search intention belongs to, acquire the relevant words to be selected in which the scene matching score exceeds a certain score (preset scene matching threshold) in the relevant recommended word library, acquire the search relevant words matched with the search keywords in the relevant words to be selected (determined according to the relevant matching score of the relevant words to be selected and the search keywords, the relevant matching score needs to exceed the preset relevant matching threshold), and return data according to a certain number (the preset scene recommended number in each preset scene). And if the search associated words matched with the search keywords are not retrieved in a certain scene, the scene is not displayed. If the number of the returned data does not reach the preset scene recommendation number in each scene, that is, all the search related words of the preset scene recommendation number are not obtained in the preset scene, all the searched words are displayed. And setting total recommended quantity, and if the recommended quantity of the sub-types does not reach the total quantity, increasing the scene recommended quantity of other scene types to ensure that the total recommended quantity is unchanged. For example, if the recommended data of the patent category is not enough, the related words of the information category or the hot search category can be returned for complement.
By customizing different search scenes, different types of associated recommended words are recalled based on different types of user search behaviors, for example, enterprise-class recall enterprise names, brands, popular apps, well-known entrepreneurs and other dimensional data, response capability within 100ms and high concurrency capability support of 500tps are achieved by calling a big data elastic search, and user search experience is improved.
205. And displaying the relevant word recommendation set with relevant scene information of relevant words.
The relevant words in the relevant word recommendation set are distinguished according to the scenes to which the relevant words belong, and are displayed in a classified manner according to the preset scene recommendation quantity, and the display is carried with the information of the scenes to which the relevant words belong; the quantity of the returned data in each preset scene can be set according to actual requirements, and the returned associated recommended words are displayed according to the recommended quantity of the preset scenes.
The preset scenes can be set, and the recommended number of the preset scenes can be set respectively for each preset scene; the display sequence of each preset scene can be set, or when a plurality of identified search intents are provided, the scenes associated with the search intents are displayed according to the ranking of the search intents (ranking according to the association degree of the search intents and the search keywords). When a setting request for a preset scene, a preset scene recommendation number or a preset scene display sequence is acquired, setting a preset scene type, a preset scene recommendation number or a preset scene display sequence.
The method comprises the following steps of presetting scenes and presetting recommended scene quantity, and setting according to actual requirements: 1) Associating a recommended word scene type; 2) The scene recommendation quantity of the associated recommended words; 3) A prefix or full match pattern; 4) Whether to return an intent parameter.
In an embodiment of the application, the whole structure of the search of the association recommended word is optimized, and for different scenes, corresponding structural data can be acquired according to a parameter interaction mode, so that customization of different scenes is realized. The specific implementation logic of parameter interaction: the back end transmits a type parameter and a size parameter to flexibly select a desired data type and data quantity, such as a parameter setting type:5, size:10, business category scene data is returned, for a total of 10 items.
According to the technical scheme of the search association recommendation method, by means of a data warehouse platform, a data center service operation platform, machine learning data cleaning, an algorithm intention recognition model, an algorithm word error correction model, an elastic search, kafka data real-time stream and python service code integration capacity under big data, the user can efficiently find the content to be searched when using a search function, the setting is made, the use experience of the search function is improved, and the flexible updating of a word stock can be met. The user search behavior may provide data support for the developer to provide model optimization. And dynamic data updating is performed according to abundant data resources on the big data side, so that timeliness and high availability of the associated recommended words are guaranteed.
The method for generating the related-word recommendation set shown in fig. 3 may implement the following steps:
301. determining a preset scene associated with the search intention.
Searching whether the search intention has an associated preset scene or not; if yes, acquiring search associated words matched with the search keywords in each preset scene associated with the search intention according to the preset scene associated with the search intention and the corresponding preset scene recommendation quantity; and if not, acquiring the search associated words matched with the search keywords in each preset scene according to all preset scenes and the corresponding preset scene recommendation quantity.
In the embodiment of the present application, a scene refers to a scene associated with search related words, and each search related word may belong to a plurality of scenes. The preset scenes comprise scenes such as enterprise scenes, patent scenes, product scenes, policy scenes, information scenes, hot searching scenes and the like; the search intent may be identified as a single intent (e.g., only enterprise category) or as multiple intents (e.g., enterprise category, product category, policy category, etc.). And if the search intention cannot be identified or the identified search intention does not belong to the preset scene type, displaying the information according to all the preset scenes.
The hot searching comprises hot searching words, and the hot searching words reflect high-frequency searching of the nearest user. Dynamic associative searches may be performed by adding hot search terms over a recent period of time (e.g., within an hour) to a remote dictionary service (redis) via a kafka real-time data stream. And the hot search type scene is used as one of the preset scenes, and if the intention is identified as the hot search type, the hot search type recommendation words are displayed according to the preset recommendation quantity.
302. And acquiring the associated word to be selected matched with the associated preset scene according to the scene relevancy information of the associated word.
When the search intention is consistent with a preset scene, acquiring the associated words to be selected according to the corresponding scene, if the search intention identifies that the enterprise is to be retrieved, belonging to an enterprise class scene; recognizing that the patent information to be searched belongs to a patent scene; the identification belongs to a hot-searching scene when the hot-searching information is required to be searched.
303. And determining the associated matching distribution of the associated words to be selected and the search keywords.
Based on a relevancy scoring algorithm (such as BM25 algorithm), a literal matching degree score, namely a relevancy matching score, of the to-be-selected related word and the search keyword based on the text is obtained.
304. And generating a relevant word recommendation set according to the relevant matching scores of the relevant words to be selected, wherein the relevant word recommendation set comprises relevant scene information of the relevant words.
And determining which to-be-selected relevant words can be matched through the scene matching score.
In the embodiment of the present application, the following steps may also be implemented:
acquiring the search heat degree and/or content comprehensive score of the associated word to be selected;
generating a comprehensive score of the associated word to be selected according to the search heat degree and/or the content comprehensive score of the associated word to be selected and the associated matching score of the associated word to be selected and the search keyword;
and sequencing the associated words to be selected according to the comprehensive score of the associated words to be selected.
In one embodiment of the application, according to the comprehensive scores, the associated words to be selected are respectively sorted according to the scenes to which the associated words belong, and a sorting result is obtained; and generating an association recommendation set of the user according to the sequencing result, and performing classification display according to the scenes to which the associated words to be selected belong.
In an embodiment of the application, the expression of the comprehensive score of the to-be-selected related word is as follows:
wherein, yscoreFor composite score, dslscoreFor associated match score, userscoreTo search for the Heat score, itemscoreFor content integration score, n represents the number of scoring dimensions.
A scoring mechanism is added in the sequencing of the associated words, high-level words can be preferentially displayed through model training content portrait data and user portrait data, and the user searching and using experience is improved. In the embodiment of the application, the comprehensive score is obtained by weighted summation of the associated matching score, the user search heat score and the content comprehensive score.
In one embodiment of the present application, the inverted arrangement may be performed by dimensions such as popularity information, length information, matching score information, and content score information of the associated word. For example, in the case of a scene type of an enterprise, the ranking may be performed according to the popularity information of the related words, the number of patents owned by the enterprise, the length information, the association matching score, the content integration score, and the like.
The method is characterized in that a sorting model is used for sorting recall results, and due to the fact that display positions are limited, the sorted results can be rearranged while accuracy and diversity of display data are considered.
In the embodiment of the present application, the following steps may also be implemented:
and generating a relevant word recommendation set according to the historical search words of the user and the search heat information of the historical search words.
In one embodiment of the present application, the following steps may be implemented:
1) Acquiring historical search terms input by a user within a preset time period.
2) And adding the historical search words into a preset historical recommended word bank as the historical words to be selected.
3) And determining the associated matching scores of the historical words to be selected and the search keywords in the preset historical recommended word bank.
4) And acquiring the searching heat degree of the history words to be selected.
5) And acquiring historical associated words with preset historical recommendation quantity from the to-be-selected historical words according to the association matching score and the search heat score of the to-be-selected historical words, wherein the scene type of the historical associated words is a historical search scene.
6) And taking the history relevant words as search relevant words and adding the search relevant words into the relevant recommendation set.
In an embodiment of the application, a data real-time stream processing mechanism is introduced, so that the latest searching behavior of a user can be dynamically displayed in the next use of the user, and the higher the searching frequency of the user is, the greater the chance of displaying the associated recommended word is. Historical input words within a set period of time, such as words input within the last three days, may be added to the historical recommended word library. The scene type of the history input word is set as a history search scene. The "history search" scene may be set independently of the preset scene type described above. When closing, searching relevant words and only displaying words in a preset scene type; when the system is started, the associative recommended words of the 'historical search' scene (or label) can be added in the whole display. And the data in the historical recommended word bank is updated in real time, and the data in the historical recommended word bank is relatively dynamic by comparing the data in the associated recommended word bank.
In an embodiment of the application, when displaying the related-word recommendation set with related-word related scene information, any one or any combination of the following manners may be implemented:
1) And respectively displaying the associated recommended words corresponding to each scene type according to the preset scene recommended quantity.
2) Acquiring historical search intents of a user during searching within a set time period, and dynamically adjusting the recommended number of scenes corresponding to the historical search intents according to the times of preset scene types associated with the historical search intents and the feedback of the user on search results; and respectively displaying the associated recommended words corresponding to the scene types according to the adjusted scene recommended quantity. For example: when the user has more historical search enterprise and patent information, the number of associated recommended words of the enterprise scene type and the patent scene type is dynamically increased under the condition of keeping the total recommended number unchanged, and the recommended numbers of other types are correspondingly reduced. The recommended number of each preset scene type can be correspondingly adjusted according to the preset scene type proportion data associated with the historical search intention.
3) Acquiring a user portrait according to user behaviors; acquiring a scene type preferred by a user according to the user portrait; dynamically adjusting the recommended number of scenes according to the scene types preferred by the user; and respectively displaying the associated recommended words corresponding to the scene types according to the adjusted scene recommended quantity. For example: the user preference is to search product information or hot search information, and under the condition that the total recommendation quantity is kept unchanged, the quantity of associated recommendation words of the product scene type and the hot search scene type is dynamically increased, and the recommendation quantity of other types is correspondingly reduced.
By the search association recommendation method, the recommended associated recommended words are classified and displayed according to scenes, and the displayed associated words or the scene types of the associated words can be dynamically adjusted according to user feedback, so that the associated recommended words recommended to the user can more accurately match with the user requirements.
As shown in a schematic diagram of a framework of a search association recommendation apparatus in fig. 4, an embodiment of the present application provides a search association recommendation apparatus, including:
a thesaurus generation module 401, configured to generate an associated recommended thesaurus, where the thesaurus includes an associated word set and scene relevancy information of different associated words;
a search request obtaining module 402, configured to obtain a search request input by a user, where the search request carries a search keyword;
an intention identification module 403, configured to perform intention identification processing on the search request to obtain a search intention;
a recommendation generation module 404, configured to generate a relevant word recommendation set associated with the search keyword and the search intention by using the relevant recommendation thesaurus;
and a recommendation display module 405, configured to display the relevant word recommendation set with relevant scene information of the relevant word.
In some possible implementations, an electronic device according to an implementation of the application may include at least one processor, and at least one storage. Wherein the storage device stores at least one computer program, and when the computer program is executed by the processor, the processor is enabled to execute the steps of the method according to the various embodiments of the present application described in the above technical solutions of the present specification.
In some possible embodiments, the various aspects of the present application may also be implemented as a computer-readable storage medium having stored thereon a computer program for implementing the steps in the method according to the various embodiments of the present application described in the above-mentioned technical solutions of the present specification when the computer program is executed by a processor of an electronic device.
The computer-readable storage medium may be, for example but not limited to: an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of the readable storage medium include: an electrical connection having one or more wires, a portable disk, 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.
Program code embodied on a readable storage medium may be transmitted using any appropriate medium, including but not limited to: wireless, wired, fiber optic, RF, etc., or any suitable combination of the foregoing.
Program code for carrying out operations of the present application may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the consumer electronic device, partly on a remote electronic device, or entirely on the remote electronic device or server. In the case of remote electronic devices, the remote electronic devices may be connected to the consumer electronic device through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external electronic device (e.g., through the internet using an internet service provider).
As will be appreciated by one skilled in the art, aspects of the present application may be embodied as a system, method or program product. Accordingly, various aspects of the present application may be embodied in the form of: an entirely hardware embodiment, an entirely software embodiment (including firmware, microcode, etc.) or an embodiment combining hardware and software aspects that may all generally be referred to herein as a "circuit," module "or" system.
The above embodiments are preferred embodiments of the present application, and the protection scope of the present application is not limited by the above embodiments, so: all equivalent changes made according to the structure, shape and principle of the present application shall be covered by the protection scope of the present application.
Claims (10)
1. A search association recommendation method, the method comprising:
generating a related recommended word bank, wherein the word bank comprises a related word set and scene relevancy information of different related words;
acquiring a search request input by a user, wherein the search request carries a search keyword;
performing intention identification processing on the search request to acquire a search intention;
generating a relevant word recommendation set associated with the search keyword and the search intention by using the relevant recommendation word bank;
and displaying the relevant word recommendation set with relevant scene information of the relevant words.
2. The search association recommendation method according to claim 1, wherein the generating of the associated recommendation thesaurus specifically comprises:
establishing an associated recommended word bank, wherein the associated recommended word bank comprises an associated word set formed by associated words;
and determining scene matching distribution of each relevant word in each preset scene according to the correlation degree of the relevant word and different preset scenes.
3. The search association recommendation method according to claim 2, wherein the determining of the scene matching score of each related word in each preset scene further comprises any of the following manners or any combination thereof:
a) Carrying out data cleaning and data enhancement processing on the associated recommended word bank to generate different scene matching scores of different associated words;
b) Generating scene matching scores for different associated words in the associated recommended word library based on the historical behaviors of the user and the associated information of the associated words;
c) And generating different scene matching scores of different associated words in the associated recommended word library through an expert system.
4. The search association recommendation method according to claim 2 or 3, further comprising: and determining the scene of each relevant word according to the scene matching distribution of different relevant words.
5. The search association recommendation method according to claim 1, wherein the generating of the associated word recommendation set associated with the search keyword and the search intention using the associated recommendation thesaurus specifically comprises:
determining a preset scene associated with the search intention;
acquiring a to-be-selected associated word matched with the associated preset scene according to the scene correlation degree information of the associated word;
determining the associated matching score of the associated word to be selected and the search keyword;
and generating a relevant word recommendation set according to the relevant matching scores of the relevant words to be selected, wherein the relevant word recommendation set comprises relevant scene information of the relevant words.
6. The search association recommendation method according to claim 5, wherein the generating of the set of related word recommendations further comprises:
acquiring the search heat degree and/or content comprehensive score of the associated word to be selected;
generating a comprehensive score of the associated word to be selected according to the search heat degree and/or the content comprehensive score of the associated word to be selected and the associated matching score of the associated word to be selected and the search keyword;
and sequencing the associated words to be selected according to the comprehensive scores of the associated words to be selected.
7. The search association recommendation method according to claim 5, wherein the generating of the associated word recommendation set associated with the search keyword and the search intention using the associated recommendation thesaurus further comprises:
and generating a relevant word recommendation set according to the historical search words of the user and the search heat information of the historical search words.
8. A search association recommendation apparatus, comprising:
the word bank generating module is used for generating a relevant recommended word bank, and the word bank comprises a relevant word set and scene relevancy information of different relevant words;
the device comprises a search request acquisition module, a search query processing module and a search query processing module, wherein the search request acquisition module is used for acquiring a search request input by a user, and the search request carries a search keyword;
the intention identification module is used for carrying out intention identification processing on the search request and acquiring a search intention;
the recommendation generation module is used for generating a relevant word recommendation set associated with the search keyword and the search intention by using the relevant recommendation word bank;
and the recommendation display module is used for displaying the relevant word recommendation set with relevant word relevant scene information.
9. An electronic device, comprising:
at least one processor;
storage means for storing at least one computer program;
the at least one computer program, when executed by the at least one processor, causes the at least one processor to implement the method of any one of claims 1 to 7.
10. A computer-readable storage medium having stored thereon a computer program, characterized in that: the computer program, when executed by a processor, implements the method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210870530.6A CN115269959A (en) | 2022-07-23 | 2022-07-23 | Search association recommendation method and device, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210870530.6A CN115269959A (en) | 2022-07-23 | 2022-07-23 | Search association recommendation method and device, electronic equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115269959A true CN115269959A (en) | 2022-11-01 |
Family
ID=83769654
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210870530.6A Pending CN115269959A (en) | 2022-07-23 | 2022-07-23 | Search association recommendation method and device, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115269959A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116150456A (en) * | 2023-04-18 | 2023-05-23 | 中信天津金融科技服务有限公司 | Intelligent archive management method, device, electronic equipment and medium |
| CN116628129A (en) * | 2023-07-21 | 2023-08-22 | 南京爱福路汽车科技有限公司 | Auto part searching method and system |
-
2022
- 2022-07-23 CN CN202210870530.6A patent/CN115269959A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116150456A (en) * | 2023-04-18 | 2023-05-23 | 中信天津金融科技服务有限公司 | Intelligent archive management method, device, electronic equipment and medium |
| CN116150456B (en) * | 2023-04-18 | 2023-08-11 | 中信天津金融科技服务有限公司 | Intelligent archive management method, device, electronic equipment and medium |
| CN116628129A (en) * | 2023-07-21 | 2023-08-22 | 南京爱福路汽车科技有限公司 | Auto part searching method and system |
| CN116628129B (en) * | 2023-07-21 | 2024-02-27 | 南京爱福路汽车科技有限公司 | Auto part searching method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110222160B (en) | Intelligent semantic document recommendation method and device and computer readable storage medium | |
| CN109635273B (en) | Text keyword extraction method, device, equipment and storage medium | |
| EP4027268A1 (en) | Training method for classification model, sample classification method and apparatus, and device | |
| US10565533B2 (en) | Systems and methods for similarity and context measures for trademark and service mark analysis and repository searches | |
| US20180268038A1 (en) | Systems and Methods for Similarity and Context Measures for Trademark and Service Mark Analysis and Repository Searches | |
| CN111144723A (en) | Method and system for recommending people's job matching and storage medium | |
| US20220012296A1 (en) | Systems and methods to automatically categorize social media posts and recommend social media posts | |
| US20100205198A1 (en) | Search query disambiguation | |
| CN112800170A (en) | Question matching method and device and question reply method and device | |
| US20110087961A1 (en) | Method and System for Assisting in Typing | |
| KR20200007969A (en) | Information processing methods, terminals, and computer storage media | |
| CN115269959A (en) | Search association recommendation method and device, electronic equipment and storage medium | |
| CN109937417A (en) | The system and method for context searchig for electronical record | |
| WO2006083939A2 (en) | Prioritization of search responses system and method | |
| CN104834651B (en) | Method and device for providing high-frequency question answers | |
| CN109947902B (en) | Data query method and device and readable medium | |
| CN111105209A (en) | Job resume matching method and device suitable for post matching recommendation system | |
| CN110968684A (en) | Information processing method, device, equipment and storage medium | |
| CN110597978B (en) | Article abstract generation method, system, electronic equipment and readable storage medium | |
| CN110008473B (en) | Medical text named entity identification and labeling method based on iteration method | |
| CN111325018B (en) | Domain dictionary construction method based on web retrieval and new word discovery | |
| CN112115232A (en) | Data error correction method and device and server | |
| US20210374276A1 (en) | Smart document migration and entity detection | |
| CN113821605A (en) | Event extraction method | |
| CN114329225A (en) | Search method, device, equipment and storage medium based on search statement |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB02 | Change of applicant information | ||
| CB02 | Change of applicant information |
Address after: 518051 2201, block D, building 1, bid section 1, Chuangzhi Yuncheng, Liuxian Avenue, Xili community, Xili street, Nanshan District, Shenzhen, Guangdong Applicant after: Qizhi Technology Co.,Ltd. Address before: 518051 2201, block D, building 1, bid section 1, Chuangzhi Yuncheng, Liuxian Avenue, Xili community, Xili street, Nanshan District, Shenzhen, Guangdong Applicant before: Qizhi Network Technology Co.,Ltd. |