Position search method and device based on search terms, electronic equipment and storage medium

Abstract

The invention relates to a position search method and device based on search terms, electronic equipment and a storage medium. The method comprises the following steps: obtaining a search term input by a job hunting user; based on a retrieval field in a preset structured retrieval condition, analyzing the search terms, and identifying the search terms corresponding to the retrieval field from the search terms; corresponding retrieval fields in the retrieval conditions are filled with the recognized retrieval words to construct the retrieval conditions; checking whether the retrieval condition meets a sufficient condition or not; responding to the condition that the retrieval condition meets a sufficient condition, and searching advertisement positions meeting the retrieval condition in an advertisement position database; and returning the searched advertisement position to the search output device. According to the embodiment of the invention, accurate retrieval is realized, the retrieved advertisement position conforms to the expectation of the user, and retrieval resources are saved.

Description

Technical Field

[0001] The present invention relates to the technical field of Internet data processing, and in particular to a job search method, device, electronic device and storage medium based on search terms. Background Art

[0002] With the development of the Internet, the recruitment platforms in the existing recruitment industry have stored a large amount of job information and resume information of job seekers, and usually provide corresponding services to job seekers and recruitment users respectively. For job seekers, the recruitment platform provides a resume uploading service to job seekers. Job seekers can create their own resumes on the platform and store them on the recruitment platform. The recruitment platform provides job search services to job seekers. Job seekers can enter search terms through the search window provided by the platform to find job information. For recruitment users, in addition to providing recruitment users with the usual resume search, screening and other services, the recruitment platform also provides some paid value-added services, such as increasing the exposure of positions. In order to distinguish them from ordinary positions posted by recruitment users, in the present invention, positions that increase exposure through payment are referred to as advertising positions.

[0003] When a job seeker enters a search term through the search window provided by the platform to search for a position, the usual processing method is to segment the search keyword (Query, hereinafter referred to as the search term) to obtain multiple entity keywords, and then perform fuzzy matching based on the keywords. In order to improve the search success rate of advertising positions, as long as one of the keywords can be matched, the matched advertising position will be output to the user. Although the current method can match some advertising positions, it is difficult to meet the accuracy requirements. For example, the search term entered by the user is "Shanghai Advertising Algorithm Engineer", and when segmenting the words, "Shanghai", "Advertisement", "Algorithm" and "Engineer" will be obtained respectively. When searching for advertising positions, the matching advertising positions related to "Shanghai" will also be recommended to job seekers, which will result in a large difference between the recommended positions and the user's needs, and a poor user experience.

[0004] Therefore, how to resolve the contradiction between job seekers' precise job search needs and one-sided search keywords to improve the accuracy of job recommendations is an urgent problem to be solved. Summary of the invention

[0005] In view of the technical problems existing in the prior art, the present invention proposes a job search method, device, electronic device and storage medium based on search terms, so as to improve the accuracy of advertising job search in the job search scenario based on search terms.

[0006] In order to solve the above technical problem, according to one aspect of the present invention, the present invention provides a job search method based on search terms, comprising the following steps:

[0007] Get the search term entered by the job seeker;

[0008] Based on the search field in the preset structured search condition, the search word is parsed, and the search word corresponding to the search field is identified from the search word;

[0009] Filling the identified search terms into corresponding search fields in the search conditions to construct the search conditions;

[0010] Checking whether the search condition satisfies a sufficient condition;

[0011] In response to the search condition satisfying a sufficient condition, searching an advertisement position database for an advertisement position that meets the search condition; and

[0012] The searched advertising positions are returned to the search output device.

[0013] Optionally, the step of parsing the search term to identify a search term corresponding to a search field from the search term includes:

[0014] Performing word segmentation and semantic analysis on the search terms input by the job seeker to obtain one or more semantic groups, each of which includes one or more entity keywords; and

[0015] The search field corresponding to each semantic group is identified, and the entity keywords in the semantic group are determined as the search terms of the corresponding search field.

[0016] Optionally, after identifying the search field corresponding to the semantic group, each entity keyword in the semantic group is matched in the field dictionary corresponding to the search field, and the dictionary vocabulary matching the entity keyword is determined as the search term of the corresponding search field.

[0017] Optionally, the search field includes a function field, and when it is identified that the search field corresponding to the semantic group is a function field, the following steps are included:

[0018] Perform fuzzy matching on semantic groups in the functional dictionary to obtain multiple candidate functional dictionary words;

[0019] Calculating a relevance score of each candidate functional dictionary word to the semantic group; and

[0020] Candidate function dictionary words whose relevance scores are greater than or equal to the threshold and / or whose rankings are before the ordinal threshold when sorted from high to low according to the relevance scores are determined as search terms for the function field.

[0021] Optionally, the job search method based on search terms further comprises the step of calculating the relevance score between each candidate functional vocabulary and the semantic group:

[0022] Performing word segmentation processing on target candidate functional dictionary words to obtain multiple functional word segments;

[0023] Compare each entity keyword in the semantic group with the function segmentation words in the target candidate function dictionary vocabulary, and use the function segmentation words that match each entity keyword in the semantic group as the target function segmentation words;

[0024] Obtain the functional score of the target functional segmentation word, and calculate the sum of the functional scores of all target functional segmentation words;

[0025] Calculating a weighted sum of the number of entity keywords in the semantic group and the number of segmented words in the target candidate functional dictionary vocabulary; and

[0026] Calculate the quotient of the sum of the functional scores of all target functional word segments and the weighted sum of the number of entity keywords in the semantic group and the number of word segments in the target candidate functional dictionary vocabulary, and use the quotient as the relevance score between the target candidate functional dictionary vocabulary and the semantic group.

[0027] Optionally, the job search method based on search terms further includes the step of constructing a job-seeking user portrait:

[0028] Obtaining behavior data of job seekers, wherein the behavior data includes behavior type and job information corresponding to the behavior;

[0029] Using users, user behaviors and positions as three-dimensional coordinate axes, three-dimensional portrait statistics of job seekers are constructed, wherein the position dimension includes a plurality of position sub-information corresponding to the search fields respectively; and

[0030] The user's intention preference for each position sub-information is calculated based on the three-dimensional portrait statistical data, wherein the user's intention preference for each position sub-information constitutes the user portrait.

[0031] Optionally, in the job search method based on search terms, before checking whether the search condition satisfies the sufficient condition, the method further comprises:

[0032] Check whether the search field in the search condition meets the filling requirement;

[0033] In response to a search field in the search condition not satisfying a filling requirement, extracting the job seeker's intention preference corresponding to the search field not satisfying the filling requirement from the job seeker's portrait data; and

[0034] The job seeker's intention preference is added as the first expanded search term to the corresponding search field.

[0035] Optionally, in the job search method based on search terms, before checking whether the search field in the search condition satisfies the sufficient condition, the method further includes:

[0036] Check whether the search field in the search condition meets the filling requirement;

[0037] In response to a search field in the search condition not satisfying a filling requirement, querying a related search word library for a related word based on a search word of the search field in the current search condition not satisfying the filling requirement; and

[0038] One or more associated words obtained from the query are added as second expanded search words to the corresponding search field.

[0039] Optionally, the step of checking whether the search condition satisfies a sufficient condition includes:

[0040] Check whether the specific search field in the current search condition contains the preset search term;

[0041] In response to the specific search field containing the preset search term, determining that the search condition satisfies the sufficient condition, and in response to the specific search field not containing the preset search term, determining that the search condition does not satisfy the sufficient condition;

[0042] Correspondingly, when the search condition does not meet the sufficient condition, the search for the advertised position is stopped.

[0043] In order to solve the above technical problems, according to another aspect of the present invention, the present invention also provides a job search device based on search terms, including a search term acquisition module, a search term parsing module, a retrieval condition module, a verification module and a search module, wherein the search term acquisition module is configured to acquire the search term input by a job seeker; the search term parsing module is configured to parse the search term based on the retrieval field in the preset structured retrieval condition, and identify the search term corresponding to the retrieval field from the search term; the retrieval condition module is configured to fill the identified search term into the corresponding retrieval field in the retrieval condition to construct the retrieval condition; the verification module is configured to at least verify whether the retrieval condition satisfies a sufficient condition; the search module is configured to search for an advertising position that meets the retrieval condition in the advertising position database in response to the retrieval condition satisfying the sufficient condition, and return the searched advertising position to the search output device.

[0044] Optionally, the search term-based job search device further includes a user portrait module and a first extension module, the user portrait module being configured to use user, user behavior and job as three-dimensional coordinate axes respectively, to construct three-dimensional stereoscopic portrait statistical data of each job seeker, and to calculate the user's intention preference for each job sub-information based on the three-dimensional stereoscopic portrait statistical data; correspondingly, the verification module is configured to verify whether the retrieval condition meets the filling requirement before verifying whether the retrieval condition meets the sufficient condition; the first extension module is configured to extract the job seeker's intention preference corresponding to the retrieval field that does not meet the filling requirement from the job seeker's portrait data in response to the retrieval condition not meeting the filling requirement; correspondingly, the retrieval condition module adds the job seeker's intention preference as the first extended search term to the corresponding retrieval field.

[0045] Optionally, the search term-based job search device further includes an associated search term library construction module and a second expansion module, wherein the associated search term library construction module is configured to construct an associated search term library corresponding to each search field; correspondingly, the verification module is configured to verify whether the search condition meets the filling requirement before verifying whether the search condition meets the sufficient condition; the second expansion module is configured to query the associated term in the associated search library based on the search term of the search field that does not meet the filling requirement in the current search condition in response to the search condition not meeting the filling requirement; correspondingly, the search condition module adds the associated term obtained by the query as the second expanded search term to the corresponding search field.

[0046] In order to solve the above technical problem, according to another aspect of the present invention, the present invention also provides an electronic device, which includes a processor and a memory, and a computer program instruction set is stored in the memory. When the processor executes the computer program instruction set on the memory, the aforementioned search term-based job search method is implemented.

[0047] According to another aspect of the present invention, the present invention further provides a computer-readable storage medium, wherein a computer program instruction set is stored on the computer-readable storage medium, and when the computer program instruction set is executed by a processor, the aforementioned job search method based on search terms is implemented.

[0048] According to another aspect of the present invention, the present invention further provides a computer program product, which includes a computer program instruction set, and when the computer program instruction set is executed by a processor, the above-mentioned job search method based on search terms is implemented.

[0049] The embodiment of the present invention realizes precise retrieval by setting structured retrieval conditions, so that the retrieved advertising positions meet the user's expectations; and by judging the sufficient conditions of the retrieval conditions, it can effectively reduce the output of advertising positions that do not meet the user's needs, reduce the trouble to the user, and save retrieval resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] The preferred embodiments of the present invention will be further described in detail below with reference to the accompanying drawings, wherein:

[0051] Figure 1 is a general flow chart of a job search method based on search terms according to an embodiment of the present invention;

[0052] Figure 2 is a flow chart of a method for searching advertising positions based on search terms according to an embodiment of the present invention;

[0053] Figure 3 is a flowchart of a process for identifying a search term corresponding to a search field from a search term according to an embodiment of the present invention;

[0054] Figure 4 is a processing flow chart of determining the search term of the functional search field according to the search term according to one embodiment of the present invention;

[0055] Figure 5 is a flow chart of a method for searching advertising positions based on search terms according to another embodiment of the present invention;

[0056] Figure 6 is a schematic diagram of statistical data of a three-dimensional image according to an embodiment of the present invention;

[0057] Figure 7 is a flowchart of constructing a function-related search vocabulary according to an embodiment of the present invention;

[0058] Figure 8 is a schematic diagram of a heterogeneous graph according to an embodiment of the present invention;

[0059] Fig. 9 is a principle block diagram of a position search device based on search terms according to an embodiment of the present invention;

[0060] Fig.10 is a principle block diagram of an advertisement position search device based on search terms according to another embodiment of the present invention;

[0061] Fig.11 is a principle block diagram of an advertisement position search device based on search terms according to another embodiment of the present invention; and

[0062] Fig.12It is a schematic diagram of the hardware structure principle of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0063] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0064] In the following detailed description, reference may be made to the various specification drawings that are part of the present application and are used to illustrate specific embodiments of the present application. In the accompanying drawings, similar reference numerals describe substantially similar components in different figures. The various specific embodiments of the present application are described below in sufficient detail so that a person of ordinary skill in the art with relevant knowledge and skills in the art can implement the technical solutions of the present application. It should be understood that other embodiments may also be used or structural, logical or electrical changes may be made to the embodiments of the present application.

[0065] Figure 1 This is a general flow chart of a job search method based on search terms according to an embodiment of the present invention. After a job seeker enters a search term through the search window provided by the recruitment platform, a job search is performed simultaneously through two processes based on the search term, wherein an advertisement job search is performed in step S10, and after the advertisement job is searched, the advertisement job is sorted through an output device in step S20, such as first performing a rough sorting, and then performing a fine sorting with more sorting conditions. In another process, an ordinary job search is performed in step S30, and then in step S40, the searched ordinary jobs are sorted, and the sorting logic is similar to that of the advertisement job sorting, such as first performing a rough sorting, and then performing a fine sorting with more sorting conditions. Finally, in step S50, the jobs obtained by the two processes are comprehensively sorted, and the sorted results are output to the user.

[0066] Figure 2 1 is a flow chart of a method for searching advertising positions based on search terms according to an embodiment of the present invention. The method for searching positions in this embodiment includes the following steps:

[0067] Step S11, obtaining the search term input by the job-seeking user.

[0068] Step S12: parsing and identifying the search words, and identifying the search words corresponding to the search field from the search words.

[0069] Step S13, filling the identified search terms into the corresponding search fields in the search conditions to construct the search conditions.

[0070] Step S14, checking whether the search condition meets the sufficient condition, if so, executing step S15, if not, ending the processing flow.

[0071] Step S15, searching the advertisement job database for advertisement jobs that meet the search criteria.

[0072] Step S16, returning the searched advertisement positions to the search output device, and ending the processing flow.

[0073] In order to accurately express job-seeking needs according to the search terms input by the user, the present invention constructs a structured search condition, which is composed of multiple search fields, as shown in Table 1 below.

[0074] Table 1

[0075]

[0076] Among them, the "Function" field is used to represent a category of positions in the recruitment platform. For example, each job posted by a recruitment user has a corresponding, unique position ID on the recruitment platform. Different position IDs correspond to different positions, and positions belonging to the same category have the same function ID on the recruitment platform. For example, recruitment users A, B, and C each posted a position for recruiting accountants, which correspond to three different position IDs on the recruitment platform, but the functions of the three positions are all "accountants" and have the same function ID.

[0077] In order to indicate the importance of each search field in the search conditions, a corresponding weight is set for each search field. In the embodiment shown in Table 1, three weight levels are represented by "high", "medium" and "low".

[0078] The search term input by the user is, for example, a piece of text. In step S12, when parsing the search term, the search term input by the job seeker is segmented and semantically analyzed to obtain one or more semantic groups, each of which includes one or more entity keywords. Based on the semantic group and the search field name, the search field corresponding to each semantic group is identified, and the entity keyword in the semantic group is determined as the search term of the corresponding search field.

[0079] Specifically, see Figure 3 , Figure 3 The present invention is a flowchart of a process for identifying a search term corresponding to a search field from a search term according to an embodiment of the present invention, which specifically includes the following steps:

[0080] Step S121, determine whether the search text includes segmentation symbols, such as spaces or punctuation marks such as commas and semicolons. When the search text includes segmentation symbols such as spaces or punctuation marks, in step S122, the search text is segmented based on the segmentation symbols, and each continuous text after segmentation is regarded as a semantic group. If there are no spaces or punctuation marks in the search text, but it is a continuous text, in step S123, a natural language processing method is used for semantic analysis, and the continuous text is segmented according to semantics to obtain multiple semantic groups. For example, the text is first segmented into sentences or words to obtain multiple text fragments, and then each text fragment is converted into a vector using a model such as BERT, and then the vector cosine similarity of adjacent text fragments is calculated, and then the semantic similarity of adjacent text fragments is judged based on the similarity, and then the two adjacent text fragments are determined based on the threshold whether to merge or be independent, and the independent text fragments finally determined are regarded as a semantic group.

[0081] Step S124: taking a semantic group as a target semantic group.

[0082] Step S125 , respectively calculating the semantic similarity between the target semantic group and each search field name.

[0083] Step S126: determine the search field name with the greatest semantic similarity as the search field corresponding to the semantic group.

[0084] Step S127, extracting entity keywords from the target semantic group.

[0085] Step S128: Use the entity keyword as the search term for the corresponding search field.

[0086] Step S129, determine whether there are semantic groups that have not been calculated, if yes, return to step S124, if no, end the processing flow.

[0087] In the aforementioned embodiment, the entity keywords extracted from the target semantic group are used as the search terms of the search field. In another specific embodiment, each entity keyword of the semantic group is matched in the field dictionary corresponding to the search field, and the dictionary vocabulary that matches the entity keyword is determined as the search term of the corresponding search field. In this embodiment, the search terms are standardized by replacing the vocabulary input by the user with the vocabulary included in the dictionary, which facilitates subsequent searches. For example, for each search field in Table 1, a corresponding field dictionary is set in the recruitment platform. For example, for the "city" search field, the official names, abbreviations, aliases, etc. of all cities in the country are included in its city dictionary. For the "welfare" search field, its welfare dictionary includes welfare expressions from all walks of life.

[0088] For the "function" search field, since this search field is the field with the highest weight in the search conditions, in one embodiment, in order to increase the exposure rate of the advertised position and meet the job-seeking needs of job seekers, it is usually necessary to determine multiple search terms when determining the search terms for this search field.

[0089] See also Figure 4 , Figure 4 The present invention is a flowchart of a process of determining a search term for a functional search field based on a search term according to an embodiment of the present invention. Specifically, the flowchart includes:

[0090] Step S121a, fuzzy matching is performed on the semantic group corresponding to the function search field in the function dictionary to obtain a plurality of candidate function dictionary words.

[0091] Step S122a, calculating the relevance score between each candidate functional dictionary word and the semantic group.

[0092] Step S123a, determining candidate function dictionary words whose relevance scores are greater than or equal to the threshold and / or whose ranking is before the threshold ordinal number when sorted from high to low according to the relevance scores as search terms for the function field.

[0093] Among them, the job dictionary stored in the recruitment platform includes all job expression words, such as "recommendation algorithm engineer", "search algorithm engineer", "advertising algorithm engineer", etc. In step S121a, multiple candidate job dictionary words corresponding to the semantic group can be obtained by fuzzy matching. For example, when the semantic group is "algorithm engineer", multiple candidate job dictionary words such as "recommendation algorithm engineer", "search algorithm engineer", "advertising algorithm engineer" and so on can be matched.

[0094] In step S122a, when calculating the relevance score between each candidate functional dictionary word and the semantic group, a variety of calculation methods can be used. For example, each candidate functional dictionary word and the semantic group corresponding to the functional search field are converted into vectors respectively, the difference between the vectors is calculated, and the vector difference is used as the relevance score. In another embodiment, the recruitment platform also stores the functional score of each functional word segmentation that constitutes a functional expression vocabulary, and the functional scores of all functional word segments constitute a functional score database. The functional word segmentation refers to a semantic unit that constitutes the functional expression vocabulary. For example, for "recommended algorithm engineer", its functional word segmentation includes "recommendation", "algorithm" and "engineer". In one embodiment, the functional score of each functional word segmentation is calculated by the following formula 1-1:

[0095]

[0096] Where i represents the sequence number of the functional word in the functional score database, idf irepresents the functional score of the ith functional word segmentation, which indicates the importance of the functional word segmentation; where N funcall Represents the total number of functional expression words in the functional dictionary, N funci Represents the number of functional expression words covered by the i-th functional word segmentation.

[0097] For example, the total number of functional expression words in the functional dictionary is 26,000, including 200 functional expression words including "recommendation" and 2,000 functional expression words including "engineer". Therefore, the functional scores corresponding to the functional words "recommendation" and "engineer" can be calculated respectively by formula 1-1.

[0098] Therefore, in one embodiment, the step of calculating the relevance score of a candidate functional vocabulary and a semantic group includes:

[0099] First, the target candidate functional dictionary vocabulary is segmented to obtain multiple functional segmented words, and the number of functional segmented words N in the target candidate functional dictionary vocabulary is determined. f .

[0100] Then, each entity keyword in the semantic group is compared with the functional word segmentation in the target candidate functional dictionary vocabulary, and the functional word segmentation matching each entity keyword in the semantic group is obtained as the target functional word segmentation, and the number of entity keywords N in the semantic group is obtained. q .

[0101] Then, the relevance score between the target candidate functional dictionary vocabulary and the semantic group is calculated according to Formula 1-2.

[0102]

[0103] Where sim represents the relevance score between the target candidate function dictionary vocabulary and the semantic group; where idf i represents the idf value of the i-th function word in the target candidate function dictionary; m i is 0 or 1, when the i-th functional segmentation word matches an entity keyword in the semantic group, m i =1 when mismatched i =0; N q Indicates the number of entity keywords in the semantic group, and the 0.7 in front of it indicates the weight; N fRepresents the number of functional word segments in the target candidate functional dictionary vocabulary, and the 0.3 in front of it represents the weight. According to formula 1-2, the numerator calculates the sum of the functional scores of the functional word segments in the target candidate functional dictionary vocabulary that match the entity keywords in the semantic group, and the denominator calculates the weighted sum of the number of entity keywords in the semantic group and the number of word segments in the target candidate functional dictionary vocabulary, where the weight of the number of entity keywords and the weight of the number of functional word segments are constants, which are respectively taken as 0.7 and 0.3 in this embodiment. However, it can be known that it can be flexibly set according to the construction of the database or the field. The relevance score of the target candidate functional dictionary vocabulary and the semantic group is obtained by calculating the quotient of the sum of the functional scores of all target functional word segments and the weighted sum of the number of entity keywords in the semantic group and the number of word segments in the target candidate functional dictionary vocabulary.

[0104] After calculating the relevance score between each candidate functional dictionary word and the semantic group in step S122a, in step S123a, in one embodiment, the relevance score is compared with a threshold, and the candidate functional dictionary words greater than or equal to the threshold are determined as search terms for the functional field, or, all candidate functional dictionary words are sorted in descending order of the relevance score, and the candidate functional dictionary words before the ordinal threshold are determined as search terms for the functional field.

[0105] See also Figure 2 As shown in Table 1, in step S14, when checking whether the search condition meets the sufficient condition, for example, it is checked whether there is a search term in the high-weighted search field. If there is a search term in the high-weighted search field, it is confirmed that the sufficient condition is met, otherwise it is confirmed that the sufficient condition is not met. For another example, it is checked whether there is a preset number of search terms in the high-weighted search field. If there are no less than 3 search terms in the high-weighted search field, it is confirmed that the sufficient condition is met. When there is no search term in the high-weighted search field, or the number of search terms is less than 3, it is confirmed that the sufficient condition is not met. When the search condition does not meet the sufficient condition, the search for the advertised position is stopped, corresponding to Figure 1 , the search for advertising positions is stopped in step S10. Therefore, for the search of the search term input by the user this time, only step S30 is executed, and the positions finally output to the user do not include advertising positions. If the sufficient condition is met, in step S15, the advertising position database is searched according to the search terms in each search field in the search condition to obtain advertising positions that meet the search condition. The advertising positions are included in the positions finally output to the user.

[0106] The present invention achieves precise retrieval by setting structured retrieval conditions and presetting corresponding retrieval fields, such as function, city, industry, company, salary expectation, etc.; by setting sufficient conditions for the retrieval conditions, it can effectively reduce the output of advertising positions that do not meet the needs of users, reduce the trouble to users, and also save retrieval resources.

[0107] Furthermore, in order to solve the problem that the search terms entered by the user are insufficient and effective and sufficient search conditions cannot be obtained, in another embodiment of the present invention, it is determined whether to expand the search conditions according to the filling status of the search fields in the search conditions. Figure 5 , Figure 5 is a flow chart of a method for searching advertising positions based on search terms according to another embodiment of the present invention. Figure 2 Step S131 and step S132 are added to the process shown in FIG. Figure 2 The same is not repeated here.

[0108] After the search condition is constructed in step S13, step S131 is executed to check whether the search field in the search condition meets the filling requirement. If yes, step S14 is executed. If not, step S132 is executed to expand the search field that does not meet the filling requirement.

[0109] The filling requirements include, for example, whether a specific search field is filled with search terms, or whether the number of search terms filled in a specific search field reaches a preset number, or whether all search fields are filled with search terms.

[0110] In order to expand the search terms in the search field, in one embodiment, the search terms are expanded using user portraits, and in another embodiment, the search terms are expanded using associated words.

[0111] The present invention constructs a job-seeking user portrait database, which stores statistical data of user portraits and user intention preferences determined based on the statistical data of user portraits, and these user intention preferences constitute the job-seeking user portrait. In one embodiment, the steps of constructing the job-seeking user portrait include:

[0112] First, the behavior data of job seekers is obtained, and the behavior data includes the time of the behavior, the behavior type, and the job information corresponding to the behavior. The behavior type is, for example, viewing, reaching out, delivering, shaking hands, etc. The job information corresponding to the behavior includes job sub-information such as job title, function, industry, region, company, etc. These job sub-information can be used as search terms in various search fields.

[0113] Then, using users, user behaviors and job information as three-dimensional coordinate axes, a three-dimensional portrait of job seekers is constructed. Figure 6 , Figure 6 It is a schematic diagram of three-dimensional portrait statistical data according to an embodiment of the present invention. The X-axis is the job seeker attribute dimension, the Y-axis is the position attribute dimension, and the Z-axis is the behavior attribute dimension. The specific attribute categories (i.e., the corresponding sub-information) in the Y-axis and Z-axis can be flexibly configured as needed. The statistical data is the time-sharing behavior frequency statistics, which can be single-dimensional or multi-dimensional statistics such as minutes, hours, and days. Figure 6 It can be seen that this embodiment models the user portrait statistical data into a three-dimensional statistical data module (cube), which supports configurable expansion of the three dimensions of X, Y, and Z, and also supports the expansion of various time dimensions. The minimum time statistical granularity can be down to the minute level. This method decouples the statistical logic and statistical dimensions very well. Developers only need to configure the required dimensions and the extraction scheme of the dimensions, and no longer need to worry about complex statistical logic. It also standardizes the output interface of the statistical data, making the upper-level algorithm application more convenient. For example, when it is necessary to count the exposure, viewing, and submission behaviors of the job-seeking user's "software development engineer" in the past three days, it is only necessary to configure the function and its extraction method (the component that implements the interface extractor) in the Y dimension.

[0114] Then, the user's intention preference for each position sub-information in the position dimension is calculated based on the three-dimensional stereoscopic image statistical data, thereby obtaining the user's preference in the structured retrieval dimension. In one embodiment, the intention preference for each position sub-information in the position information is calculated by the following formulas 1-3.

[0115]

[0116] Where p represents the intention preference value for a position sub-information on the Y axis; w a is the weight of action a. For example, the weight of being exposed to a job but not clicking on it is -0.2, the weight of viewing a job is 1, and the weight of submitting a job is 2. i is the number of times the user has acted on the i-th job sub-information, N all is the number of all user behaviors, idf i is the functional score of the i-th position sub-information. The present invention calculates a corresponding functional score (ie, idf value) for each position sub-information on the Y axis, and the calculation method is shown in the above formula 1-1.

[0117] According to formula 1-3, the intention preference value of a job seeker for each position sub-information can be obtained. For example, for the position sub-information "function", the intention preference value p of each function can be calculated according to all the specific function data involved in the user's behavior over a period of time through formula 1-3. In one embodiment, all functions are sorted from high to low according to their intention preference values ​​to obtain a function ranking table, and similarly, an industry ranking table, a city ranking table, and so on are obtained. These ranking tables constitute a user portrait. In another embodiment, only the function with the highest intention preference value p is used as the function preference, and similarly, other preferences such as "industry" intention preference and "city" intention preference over a period of time are obtained.

[0118] In step S131, when it is checked that the search fields in the search conditions do not meet the filling requirements, for example, when it is checked that some search fields shown in Table 1 do not have search terms, such as the lack of search terms in the "city" search field, then in step S132, the user's portrait data is queried and the intention preference corresponding to "city" is used as the search term for the "city" search field.

[0119] In addition, in order to expand the recall volume, there are certain quantity requirements for the search terms in the search fields in the search conditions. For example, the search terms in the "function" search field and the "industry" search field are set to be no less than three. When the number of search terms obtained based on the search terms entered by the user does not meet this condition, it is necessary to expand them. In the aforementioned embodiment, the intention preference in the user portrait is used for expansion. In another expansion method, an associated search word library is used for expansion. An associated search word library corresponding to the search terms of specific search fields is stored in the database of the recruitment platform, such as the aforementioned function-associated search word library and industry-associated search word library, and other associated search word libraries may also be included, such as skill-associated search word libraries, etc.

[0120] See also Figure 7 , Figure 7 The flowchart of constructing a function-related search word library according to an embodiment of the present invention includes the following steps:

[0121] Step S701, obtain the behavior data of each job seeker in the job seeker user group, such as the job seeker's browsing of positions, posting, shaking hands with recruiters, etc. The behavior data records the time of the behavior, the behavior category, and the position corresponding to the behavior.

[0122] Step S702, obtaining the behavior type, behavior weight, and the position and function information corresponding to the behavior from the job seeker's behavior data.

[0123] Step S703, using users, positions and function information as nodes, a heterogeneous graph is constructed according to the user's behavior on the position and the corresponding relationship between the position and function information. Figure 8 , Figure 8 : is a schematic diagram of a heterogeneous graph according to an embodiment of the present invention. The circular nodes represent users, U1, U2, etc. represent user IDs, the square nodes represent specific positions sent by recruiting users, J1, J2, etc. represent position IDs, the triangular nodes represent functions, and F1, F2, etc. represent function IDs. The line from the user to the position represents the user's behavior, and includes specific behaviors and their weights.

[0124] Step S704, traversing the heterogeneous graph in the order of user, position, function, position, and user as a walking path to obtain multiple node sequences.

[0125] Step S705, each node sequence is input as a sample to the Skip-gram model. In this embodiment, the Skip-gram model is trained to obtain the high-dimensional vector corresponding to each function according to the input node sequence, and generate a similarity matrix between functions through vector retrieval. Therefore, in this embodiment, the functions and their high-dimensional vectors constitute a function-related retrieval vocabulary. Although the model in this embodiment adopts the Skip-gram model, it can be known that other Word2Vec models, such as the cbow model, can also be used.

[0126] When the function information in the above steps is replaced with other specific information, other related search terms can also be obtained according to the above method of the function related search terms, such as industry, welfare, etc. In one embodiment, at least three search terms are extended for the function search field and the industry search field. Figure 5 In step S131, it is checked whether the search terms of the function search field and the industry search field reach 3 or more. If not, based on the current search terms, the function-related search word library and the industry-related search word library are searched for the two most relevant related words, and the two most relevant related words obtained from the search are added as extended search terms to the corresponding search fields.

[0127] In this embodiment, by expanding the search terms of the specific search field, the recall rate of the advertised positions is improved and the searched advertised positions are relatively accurately matched to the user's expectations.

[0128] On the other hand, see Fig. 9 , Fig. 91 is a principle block diagram of a job search device based on search terms according to an embodiment of the present invention. The job search device based on search terms provided in this embodiment includes an interactive module 10, an advertisement job search device 20, a first search output device 30, a common job search device 40, a second search output device 50, and a comprehensive output processing device 60. The interactive module 10 is connected to the advertisement job search device 20, the common job search device 40, and the comprehensive output processing device 60, respectively. The interactive module 10 interacts with the user, and when receiving the search terms input by the user, it will be sent to the advertisement job search device 20 and the common job search device 40, respectively, and the search results will be received from the comprehensive output processing device 60 and output to the user.

[0129] The advertising position search device 20 includes a search term acquisition module 201, a search term analysis module 202, a search condition module 203, a verification module 204 and a search module 205. The search term acquisition module 201 is connected to the interaction module 10, obtains the search term input by the job seeker from the interaction module 10, and sends it to the search term analysis module 202. The search term analysis module 202 analyzes the search term based on the search field in the preset structured search condition, and identifies the search term corresponding to the search field from the search term.

[0130] The retrieval condition module 203 is connected to the database and the search term parsing module 202 respectively, obtains the retrieval field corresponding to the retrieval condition from the database, receives the search term sent by the search term parsing module 202, fills the identified search term into the corresponding retrieval field in the retrieval condition to construct the retrieval condition, and sends the retrieval condition to the verification module 204.

[0131] The verification module 204 is connected to the retrieval condition module 203 and the search module 205 respectively, and verifies whether the retrieval condition satisfies the sufficient condition according to the sufficient condition configured in the database, and sends it to the search module 205 when the sufficient condition is satisfied.

[0132] The search module 205 receives the search condition, searches for advertisement positions that meet the search condition in the advertisement position database, and returns the searched advertisement positions to the first search output device 30 .

[0133] The first search output device 30 is used to perform corresponding processing on the advertising positions searched by the search module 205, such as rough sorting, fine sorting, etc., and send the advertising positions that meet the output conditions to the comprehensive output processing device 60.

[0134] The general position search device 40 receives the search words sent by the interactive module 10, processes them according to a certain processing logic, searches for general positions in the general position database, and sends them to the second search output device 50. The second search output device 50 is similar to the first search output device 30, and performs corresponding processing on the general positions searched by the general position search device 40, such as rough sorting, fine sorting, etc., and sends the general positions that meet the output conditions to the comprehensive output processing device 60.

[0135] The comprehensive output processing device 60 performs comprehensive processing on the received advertisement positions and ordinary positions to obtain the final search results, and outputs them to the interactive module 10 .

[0136] Fig.10 is a schematic block diagram of a device for searching advertising positions based on search terms according to another embodiment of the present invention. In this embodiment, the device for searching advertising positions based on search terms 20 is configured to Fig. 9 The embodiment shown further includes a user portrait module 206 and a first expansion module 207. The user portrait module 206 constructs three-dimensional portrait statistics of each job seeker using data in the database and stores the statistics in the database.

[0137] Correspondingly, the checking module 204 checks whether the search condition meets the filling requirement before checking whether the search condition meets the sufficient condition, and sends a notification to the first expansion module 207 when it is checked that the search condition does not meet the filling requirement.

[0138] The first extension module 207 is connected to the verification module 204 and the search condition module 203 respectively, and after receiving the notification sent by the verification module 204, in response to the search condition not satisfying the filling requirement, the job seeker's intention preference corresponding to the search field is extracted from the job seeker's portrait data; the job seeker's intention preference is sent as the first extended search term to the search condition module 203, and the search condition module 203 adds the first extended search term to the corresponding search field.

[0139] Fig.11 FIG. 2 is a block diagram of a device for searching advertising positions based on search terms according to another embodiment of the present invention. Fig. 9 The embodiment shown further includes an associated search word library construction module 208 and a second expansion module 209. The associated search word library construction module 208 uses the data in the database to construct an associated search word library corresponding to each search field and stores it in the database.

[0140] Correspondingly, the checking module 204 checks whether the search condition meets the filling requirement before checking whether the search condition meets the sufficient condition, and sends a notification to the second expansion module 209 when it is checked that the search condition does not meet the filling requirement.

[0141] The second extension module 209 is connected to the verification module 204 and the search condition module 203 respectively. After receiving the notification sent by the verification module 204, in response to the search condition not satisfying the filling requirement, the second extension module 209 searches for associated words in the associated search word library based on the search words of the search field in the current search condition; sends the associated words obtained by the search as second extended search words to the search condition module 203, and the search condition module 203 adds the second extended search words to the corresponding search field.

[0142] Of course, you can also Fig.10 and Fig.11 The modules in Fig. 9 The system includes a user portrait module 206, a first expansion module 207, an associated search word library construction module 208 and a second expansion module 209, which are used to fully expand the search conditions, thereby improving the recall rate of advertising positions and making the recalled advertising positions more in line with user needs.

[0143] In another aspect, the present invention further provides an electronic device, see Fig.12 , Fig.12 It is a schematic diagram of the hardware structure principle of an electronic device according to an embodiment of the present invention. The electronic device can be implemented as a server or other various terminal devices, such as a desktop personal computer, a tablet computer, a laptop computer, a mobile phone, etc., which includes a processor 601 and a memory 602. The memory 602 stores a program instruction set. When the processor 601 executes the program instruction set on the memory 602, the aforementioned search term-based job search method is implemented.

[0144] Specifically, the processor 601 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiment of the present invention.

[0145] The memory 602 may include a large capacity memory for data or instructions. By way of example and not limitation, the memory 602 may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive or a combination of two or more of these. In appropriate cases, the memory 602 may include a removable or non-removable (or fixed) medium. In appropriate cases, the memory 602 may be inside or outside the integrated gateway disaster recovery device. In a specific embodiment, the memory 602 is a non-volatile solid-state memory.

[0146] The memory may include a read-only memory (ROM), a random access memory (RAM), a magnetic disk storage medium device, an optical storage medium device, a flash memory device, an electrical, optical or other physical / tangible memory storage device. Therefore, generally, the memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., a memory device) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the job search method based on search terms provided by the present invention.

[0147] In one example, the electronic device may further include a communication interface 603 and a bus 604. The processor 601, the memory 602, and the communication interface 603 are connected via the bus 604 and communicate with each other.

[0148] The communication interface 603 is mainly used to implement communication between various modules, devices, units and / or equipment in the embodiment of the present invention.

[0149] Bus 604 includes hardware, software or both, and the components of online data flow billing equipment are coupled to each other. For example, but not limitation, bus may include accelerated graphics port (AGP) or other graphics bus, enhanced industrial standard architecture (EISA) bus, front side bus (FSB), hypertransport (HT) interconnection, industrial standard architecture (ISA) bus, infinite bandwidth interconnection, low pin count (LPC) bus, memory bus, micro channel architecture (MCA) bus, peripheral component interconnection (PCI) bus, PCI-Express (PCI-X) bus, serial advanced technology attachment (SATA) bus, video electronics standard association local (VLB) bus or other suitable bus or two or more of these combinations. In appropriate cases, bus 604 may include one or more buses. Although the embodiment of the present invention describes and shows a specific bus, the present invention considers any suitable bus or interconnection.

[0150] The present invention also provides a computer-readable storage medium having computer program instructions stored thereon, which can be executed by a processor to implement any of the job search methods based on search terms in the aforementioned embodiments. The computer-readable storage medium can be any medium that can tangibly contain or store computer-executable instructions for use by or in combination with instruction execution systems, devices and equipment. The storage medium can be a transient computer-readable storage medium or a non-transient computer-readable storage medium. Non-transient computer-readable storage media may include, but are not limited to, magnetic storage devices, optical storage devices and / or semiconductor storage devices. Examples of such storage devices include, for example, magnetic disks, optical disks based on CD, DVD or Blu-ray technology, and persistent solid-state memories such as flash memory, solid-state drives, and the like.

[0151] The present invention also provides a computer program product, which includes a set of computer program instructions, and when the set of computer program instructions is executed by a processor, any one of the job search methods based on search terms in the above embodiments is implemented. The computer program product includes but is not limited to application installation packages published on websites and application stores, application plug-ins, and small programs that can be run in certain applications.

[0152] The above embodiments are only used to illustrate the present invention, but not to limit the present invention. Ordinary technicians in the relevant technical field can make various changes and modifications without departing from the scope of the present invention. Therefore, all equivalent technical solutions should also fall within the scope of the present invention.

Claims

1. A job search method based on search terms, characterized in that: include: Get the search term entered by the job seeker; Based on the search field in the preset structured search condition, the search word is parsed, and the search word corresponding to the search field is identified from the search word; Filling the identified search terms into corresponding search fields in the search conditions to construct the search conditions; Checking whether the search condition satisfies a sufficient condition; In response to the search condition satisfying a sufficient condition, searching an advertisement position database for an advertisement position that meets the search condition; as well as The searched advertising positions are returned to the search output device.

2. The method for searching positions based on search terms according to claim 1, characterized in that: The step of parsing the search words and identifying the search words corresponding to the search field from the search words includes: Performing word segmentation and semantic analysis on the search terms input by the job seeker to obtain one or more semantic groups, each of which includes one or more entity keywords; and The search field corresponding to each semantic group is identified, and the entity keywords in the semantic group are determined as the search terms of the corresponding search field.

3. The method for searching positions based on search terms according to claim 2, characterized in that: After identifying the search field corresponding to the semantic group, each entity keyword in the semantic group is matched in the field dictionary corresponding to the search field, and the dictionary words matching the entity keywords are determined as the search terms corresponding to the search field.

4. The method for searching positions based on search terms according to claim 2, characterized in that: The search field includes a function field, and when it is identified that the search field corresponding to the semantic group is a function field, the following steps are performed: Perform fuzzy matching on semantic groups in the functional dictionary to obtain multiple candidate functional dictionary words; Calculating a relevance score of each candidate functional dictionary word to the semantic group; and Candidate function dictionary words whose relevance scores are greater than or equal to the threshold and / or whose rankings are before the ordinal threshold when sorted from high to low according to the relevance scores are determined as search terms for the function field.

5. The method for searching positions based on search terms according to claim 4, characterized in that: Furthermore, the step of calculating the relevance score between each candidate functional vocabulary and the semantic group includes: Performing word segmentation processing on target candidate functional dictionary words to obtain multiple functional word segments; Compare each entity keyword in the semantic group with the function segmentation words in the target candidate function dictionary vocabulary, and use the function segmentation words that match each entity keyword in the semantic group as the target function segmentation words; Obtain the functional score of the target functional segmentation word, and calculate the sum of the functional scores of all target functional segmentation words; Calculating a weighted sum of the number of entity keywords in the semantic group and the number of segmented words in the target candidate functional dictionary vocabulary; and Calculate the quotient of the sum of the functional scores of all target functional word segments and the weighted sum of the number of entity keywords in the semantic group and the number of word segments in the target candidate functional dictionary vocabulary, and use the quotient as the relevance score between the target candidate functional dictionary vocabulary and the semantic group.

6. The method for searching positions based on search terms according to claim 1, characterized in that: Further steps include building a job search user profile: Obtaining behavior data of job seekers, wherein the behavior data includes behavior type and job information corresponding to the behavior; Using users, user behaviors and positions as three-dimensional coordinate axes, three-dimensional portrait statistics of job seekers are constructed, wherein the position dimension includes a plurality of position sub-information corresponding to the search fields respectively; and The user's intention preference for each position sub-information is calculated based on the three-dimensional portrait statistical data, wherein the user's intention preference for each position sub-information constitutes the user portrait.

7. The method for searching positions based on search terms according to claim 1, characterized in that: Before checking whether the search condition satisfies the sufficient condition, the method further comprises: Check whether the search field in the search condition meets the filling requirement; In response to a search field in the search condition not satisfying a filling requirement, extracting the job seeker's intention preference corresponding to the search field not satisfying the filling requirement from the job seeker's portrait data; and The job seeker's intention preference is added as the first expanded search term to the corresponding search field.

8. The method for searching positions based on search terms according to claim 1, characterized in that: Before checking whether the search field in the search condition satisfies the sufficient condition, the method further comprises: Check whether the search field in the search condition meets the filling requirement; In response to a search field in the search condition not satisfying a filling requirement, querying a related search word library for a related word based on a search word of the search field in the current search condition not satisfying the filling requirement; and One or more associated words obtained from the query are added as second expanded search words to the corresponding search field.

9. The method for searching positions based on search terms according to claim 1, characterized in that: The step of checking whether the search condition satisfies the sufficient condition comprises: Check whether the specified search field in the current search condition contains the preset search term; and In response to the specific search field containing the preset search term, determining that the search condition satisfies the sufficient condition, and in response to the specific search field not containing the preset search term, determining that the search condition does not satisfy the sufficient condition; Correspondingly, when the search condition does not meet the sufficient condition, the search for the advertised position is stopped.

10. A job search device based on search terms, characterized in that: include: A search term acquisition module configured to acquire search terms input by job seekers; A search term parsing module, configured to parse the search term based on a search field in a preset structured search condition, and identify a search term corresponding to the search field from the search term; A search condition module configured to fill the identified search terms into corresponding search fields in the search condition to construct the search condition; a verification module configured to at least verify whether the search condition satisfies a sufficient condition; as well as The search module is configured to search for advertisement positions that meet the search conditions in the advertisement position database in response to the search conditions satisfying the sufficient conditions, and return the searched advertisement positions to the search output device.

11. The job search device based on search terms according to claim 10, characterized in that: Further including: A user portrait module, which is configured to use users, user behaviors and positions as three-dimensional coordinate axes, respectively, to construct three-dimensional portrait statistics of each job seeker, and to calculate the user's intention preference for each position sub-information based on the three-dimensional portrait statistics; Correspondingly, the verification module is configured to verify whether the search condition satisfies the filling requirement before verifying whether the search condition satisfies the sufficient condition; as well as The first expansion module is configured to extract the job seeker's intention preference corresponding to the search field that does not meet the filling requirement from the job seeker's portrait data in response to the search condition not meeting the filling requirement; correspondingly, the search condition module adds the job seeker's intention preference as the first expanded search term to the corresponding search field.

12. The job search device based on search terms according to claim 10, characterized in that: Further including: an associated search word library construction module configured to construct an associated search word library corresponding to each search field; Correspondingly, the verification module is configured to verify whether the search condition satisfies the filling requirement before verifying whether the search condition satisfies the sufficient condition; as well as The second extension module is configured to query the associated words in the associated search library based on the search words of the search field that does not meet the filling requirements in the current search conditions in response to the search conditions not meeting the filling requirements; correspondingly, the search condition module adds the associated words obtained by the query as the second extended search words to the corresponding search field.

13. An electronic device, comprising a processor and a memory, wherein a computer program instruction set is stored in the memory, characterized in that: When the processor executes the computer program instruction set on the memory, the job search method based on search terms described in any one of claims 1 to 9 is implemented.

14. A computer-readable storage medium, wherein: The computer-readable storage medium stores a computer program instruction set, wherein the computer program instruction set, when executed by a processor, implements the job search method based on search terms as described in any one of claims 1 to 9.

15. A computer program product comprising a set of computer program instructions, characterized in that: When the computer program instruction set is executed by a processor, the method for searching a job based on search terms described in any one of claims 1 to 9 is implemented.

Images