A standardized method and device for talent capability labeling
By logically disassembling, standardizing and recombining the initial labels in natural language, the problem of inaccurate matching caused by natural language non-standardization in intelligent recruitment is solved, and the standardization and accurate matching of applicants' ability labels are achieved.
Patent Information
- Application Number
- CN202411810454.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-10
- Publication Date
- 2025-04-01
AI Technical Summary
The prior art is difficult to accurately match the applicant's ability in intelligent recruitment, mainly due to the non-standardization of natural language, which leads to inaccurate matching.
By obtaining the initial tag from natural language, logically disassemble and generate sub-labels, determine the sub-label type and normalize it, and finally recombining the standardized sub-labels based on the logical relational operator to generate a standardized tag combination.
The standardization of the applicant's ability labels is achieved, providing an accurate basis for subsequent matching and improving the accuracy of matching.
Smart Images
Figure CN119739864A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of Internet technology, and in particular to a method and device for standardizing talent capability labels. Background Art
[0002] With the development of mobile Internet, the application scope of intelligent recruitment is becoming more and more extensive, and with it comes the need to accurately characterize the abilities of applicants. In the job requirements and resumes of applicants, unstructured natural language is often used.
[0003] The non-standardization of natural language includes many aspects, including semantic ambiguity, non-uniform types of numerical expressions, and unstructured state expressions. These non-standard natural languages can usually be processed manually, but they are prone to inaccurate matching in intelligent recruitment.
[0004] Based on this, a more accurate standardized solution for talent capability labeling is needed. Summary of the invention
[0005] The embodiments of this specification provide a method and device for standardizing talent ability labels to solve the following technical problem: a more accurate talent ability label standardization solution is needed.
[0006] To solve the above technical problems, one or more embodiments of this specification are implemented as follows:
[0007] In a first aspect, the embodiments of this specification provide a more accurate method for standardizing talent capability labels, including:
[0008] Acquire an initial label from a natural language; logically decompose the initial label to generate multiple sub-labels, and determine the logical relationship operators between the multiple sub-labels; determine the type of the sub-label, standardize according to the type of the sub-label, and generate standardized sub-labels, wherein the type of the sub-label includes character type, numeric type or enumeration type; recombining the standardized sub-labels according to the logical relationship operator to generate a standardized label combination.
[0009] In a second aspect, one or more embodiments of this specification provide an electronic device, including:
[0010] at least one processor; and,
[0011] a memory communicatively connected to the at least one processor; wherein,
[0012] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to the first aspect.
[0013] In a third aspect, an embodiment of the present specification provides a non-volatile computer storage medium storing computer executable instructions. When a computer reads the computer executable instructions in the storage medium, the instructions enable one or more processors to execute the method described in the first aspect.
[0014] At least one of the above-mentioned technical solutions adopted in one or more embodiments of the present specification can achieve the following beneficial effects: by obtaining an initial label from a natural language; logically decomposing the initial label to generate multiple sub-labels, and determining the logical relationship operator between the multiple sub-labels; determining the type of the sub-label, standardizing according to the type of the sub-label, and generating a standardized sub-label, wherein the type of the sub-label includes a character type, a numeric type, or an enumeration type; according to the logical relationship operator, recombining the standardized sub-labels to generate a standardized label combination, thereby realizing the standardization of labels for applicants and providing an accurate basis for subsequent matching. BRIEF DESCRIPTION OF THE DRAWINGS
[0015] In order to more clearly illustrate the embodiments of this specification or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.
[0016] Figure 1 A flowchart of a method for standardizing talent capability labels provided in an embodiment of this specification;
[0017] Figure 2 A schematic diagram of a process for obtaining a final standardized label and an abnormal label from an initial label provided in an embodiment of this specification;
[0018] Figure 3 A schematic diagram of the structure of an electronic device provided in an embodiment of this specification. DETAILED DESCRIPTION
[0019] The embodiments of this specification provide a method, apparatus, device and storage medium for standardizing talent capability labels.
[0020] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this application, not all of them. Based on the embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this application.
[0021] like Figure 1 As shown, Figure 1 A flowchart of a method for standardizing talent capability labels provided in an embodiment of this specification.
[0022] Figure 1 The process in may include the following steps:
[0023] S102: Obtain initial labels from natural language.
[0024] Natural language is generally the description in the resume provided by the applicant. For example, the resume mentions that the applicant received undergraduate education in major B at University A from September 2015 to July 2019, and received graduate education in major B at University C from September 2019 to July 2022; the resume mentions that the applicant worked on algorithm research at a certain company from August 2022 to August 2023; or the resume mentions that the applicant has mastered the operation and management capabilities of CRM (customer relationship management system), etc.; the applicant's current status is "in service" and "looking for new opportunities", etc.
[0025] Based on this, the natural language in the resume can be semantically analyzed to identify the initial label. The initial label can be identified and subdivided based on a paragraph of text. For example, in the resume of the above-mentioned talent, the corresponding three paragraphs of text can be identified based on semantic analysis.
[0026] Then, by classifying these three paragraphs, the types of initial labels may include "school attended", "project experience", "professional ability", etc. For example, if the initial label is "I received undergraduate education in major B at University A from September 2015 to July 2019, and I received postgraduate education in major B at University C from September 2019 to July 2022", its type is "school attended".
[0027] Furthermore, for each type of the initial tag, a corresponding classification code may be given to facilitate subsequent searches, and the type of the initial tag may also be used as a test type for subsequent integrity checks.
[0028] Table 1: Types of initial labels and corresponding classification codes
[0029] Classification number Category code Initial label type (inspection type) 99 IGNORED Ignore Tags 101 JD_SCHOOL School 102 JD_SCHOOLLEVEL Type of school 103 JD_PROJECT Project Experience 104 JD_INDUSTRY Industry Background 105 JD_MAJOR Major 106 JD_MARRIAGE Marital status 107 JD_SEXUAL gender 108 JD_EXPERIENCE Years of working experience
[0030] As shown in Table 1, Table 1 exemplifies several types of initial labels and their corresponding classification codes. Specifically, when classifying the initial labels, the type of initial labels can be determined based on the segmented text in the following ways: 1. Classification by semantic classification algorithm; 2. Classification based on the fields in the first half of the text; 3. Windowing the context of the text, for example, dividing the initial label into 5-word text windows, performing semantic analysis, and judging the classification based on the semantic analysis results.
[0031] S104, logically decompose the initial tag to generate multiple sub-tags, and determine the logical relationship operators between the multiple sub-tags.
[0032] The logic decomposition here refers to decomposing the multiple relationships contained in the initial tag to determine the logical relationship between the multiple relationships. Each relationship corresponds to a sub-tag. The logical relationship may be parallel, inclusive, sequential, comparative, etc. The logical relationship between multiple sub-tags can be obtained through logical keywords or semantic analysis contained in the text. That is, the logical relationship keywords contained in the initial tag can be determined, and the corresponding logical relationship operators can be determined according to the logical relationship keywords.
[0033] For example, the initial label is disassembled to find out the "and, or, one of, and" contained therein, or the "3-5 years" containing the time relationship, etc. Then the corresponding logical relationship operators are determined: OPERATOR_GT (greater than), OPERATOR_LT (less than), OPERATOR_EQUAL (equal to), OPERATOR_NOT (not equal to), OPERATOR_IN (included), OPERATOR_AND (and), OPERATOR_OR (or), etc.
[0034] For an initial label "Received undergraduate education in Major B at University A from September 2015 to July 2019, and received graduate education in Major B at University C from September 2019 to July 2022", it can be queried that the logical keyword contained therein is "and", thereby determining that the corresponding multiple sub-labels may be "Major B at University A", "Major B at University C", etc., and the corresponding logical relationship symbol is OPERATOR_AND (and).
[0035] S106, determining the type of the sub-tag, and performing standardization according to the type of the sub-tag to generate a standardized sub-tag, wherein the type of the sub-tag includes a character type, a numeric type, or an enumeration type.
[0036] As mentioned above, the subtag is decomposed from the initial tag and contains multiple types. The type of the subtag here is different from the type of the initial tag, which specifically includes character type, numeric type or enumeration type.
[0037] The character-type subtag is a subtag containing specific characters, and the specific characters may be pre-specified characters with specific semantics (usually specifying relevant learning, work and other related experiences) such as "school", "company" or "project".
[0038] For character sub-tags, the standardized sub-tag corresponding to the character can be queried through the standardized dictionary. If it exists, the query result is determined as the standardized sub-tag. Otherwise, the sub-tag is embedded in the vector and the relevant standardized sub-tag is queried in the knowledge graph vector library based on the vector.
[0039] For example, for a character-type sub-tag "received undergraduate education in major B at university A", the standardized sub-tags corresponding to the characters "school" and "major" can be queried through a standardized dictionary to indicate the corresponding "education level", "school attended" and "major" and so on.
[0040] Furthermore, for some character-type sub-tags, the corresponding values may not be found in the pre-defined standardized dictionary. In this case, the knowledge model can be used to embed and vectorize the tags, and then based on the vector, similar or close vectors can be queried in the knowledge graph vector library. If a related vector with a high similarity is obtained, the related vector obtained by the query can be used to determine the standard value of the character-type sub-tag.
[0041] Numeric subtags are subtags that contain numeric values. Numeric values can be decimal places, Chinese characters, scientific notation, or values converted from text calculations. For example, "3.5k", "three years", "5*10^5", and the value "2 years" converted from the text "from 2021 to 2023".
[0042] For these different types of values, it is necessary to determine the type of the value, and then convert the value into a standard value according to the type of the value; the standard value is determined as a standardized sub-tag. For example, the aforementioned "3.5k", "three years", "5*10^5" and "2 years" are uniformly converted into the standard formats "3500", "3", "500000" and "2". In the numeric sub-tag, the specific standard format can be pre-defined based on the actual calculation needs.
[0043] Enumerated subtags are predefined subtags with multiple discrete states. For example, a person's work status may be "employed", "resigned" or "unemployed"; their job-seeking intention may be "active job-seeking", "passive job-seeking" or "wait and see". For a candidate, based on the combination of work status and job-seeking intention, there may be 6 discrete states, and these 6 discrete states can be predetermined with given values. For another example, a person's marital status may include discrete states such as "married", "unmarried" or "divorced".
[0044] When standardizing enumerated sub-tags, it is necessary to analyze the semantics of the sub-tags; call preset classification rules, and determine the standardized enumerated tags corresponding to the semantics according to the classification rules.
[0045] For example, by analyzing the semantics of "working, looking for new opportunities", we can know that the classification rule corresponding to "working status + psychological state" needs to be called. Then, it is converted into a standardized enumeration sub-tag of "type = 3" to represent the state of "working + waiting"; and for "just graduated, looking for a job", it is converted into a standardized enumeration sub-tag of "type = 4" to represent the state of "not working + actively seeking a job".
[0046] S108: Recombining the standardized sub-labels according to the logical relationship operator to generate a standardized label combination.
[0047] The recombining here refers to the recombining of multiple subtags contained in an initial tag. Generally speaking, character subtags or enumeration subtags and numeric subtags can be recombined to obtain a standardized tag combination.
[0048] The combination here includes the combination of sub-tags in order and in logic. The combination in order means that the combination order of sub-tags should conform to the preset semantic order, and the combination in logic means that the combination of sub-tags is calculated based on the logic relationship operator obtained by the above splitting.
[0049] For example, for an initial label related to an educational experience, the logical relationship symbol obtained by splitting may include "OPERATOR_AND", and the sub-labels obtained by splitting may include characters such as "JD_SCHOOL=A", "4", "JD_SCHOOL=B", and "3". Then, the standardized label combination obtained by the final combination may be "D_SCHOOL=A, 4" AND "JD_SCHOOL=B, 3".
[0050] The method obtains an initial label from a natural language; logically decomposes the initial label to generate multiple sub-labels, and determines the logical relationship operators between the multiple sub-labels; determines the type of the sub-label, standardizes the sub-label according to the type of the sub-label, and generates standardized sub-labels, wherein the type of the sub-label includes character type, numeric type or enumeration type; and recombines the standardized sub-labels according to the logical relationship operator to generate a standardized label combination, thereby achieving standardization of labels for applicants and providing an accurate basis for subsequent matching.
[0051] In one embodiment, after obtaining the standardized sub-tags and before generating the standardized tag combination, the standardized sub-tags may be post-processed. The post-processing includes deduplication, pruning, reclassification, etc. Deduplication refers to removing multiple repeated sub-tags, pruning refers to splitting sub-tags longer than a certain length, and reclassification refers to reconfirming the initial tag type corresponding to the sub-tag according to the semantics of the sub-tag.
[0052] After the post-processing is completed, the inspection type of the standardized sub-tag can be determined. Since the sub-tags are separated from the initial tags, the type of the initial tag is the inspection type of the standardized sub-tag. For the reclassified sub-tags, the initial tag type corresponding to the reclassification should be used as the inspection type of the sub-tag.
[0053] In one embodiment, since the reclassification of sub-labels and the inspection type are directly related, there should be accurate standards for whether the sub-labels should be reclassified and to which category they should be reclassified. Based on this, the reclassification of sub-labels can be carried out in the following manner: obtain the vector λ1 corresponding to the sub-label, and obtain the remaining sub-labels λ2 that have a logical relationship with the sub-label in the first initial label to generate a first label combination, the vector corresponding to the first label combination is λ3, and the ratio of [λ1+λ2] / λ3 is determined, where [λ1+λ2] represents the concatenation of the two vectors λ1 and λ2; at the same time, obtain the remaining sub-labels λ4 that have a logical relationship with the sub-label in the second initial label to generate a second label combination, the vector corresponding to the second label combination is λ5, and the ratio of [λ1+λ4] / λ5 is determined; if
[0054] Where K is a preset threshold constant, the sub-label is reclassified into the second initial label, and the second label combination is generated with the remaining sub-labels λ4, and << represents much less than, usually more than 10 times. In this way, it can be confirmed that the semantics of the sub-label is closer to the second label combination than the first label combination and the second label combination. Therefore, the sub-label should be reclassified into the second initial label, and the second label combination is generated with the remaining sub-labels λ4 to achieve more accurate reclassification.
[0055] In addition, if a standardized label combination cannot be generated, the initial label is determined as an abnormal label and stored in the abnormal processing memory. Figure 2 As shown, Figure 2 A flowchart of obtaining a final standardized label and an abnormal label from an initial label is provided in an embodiment of this specification.
[0056] In the second aspect, based on the same idea, one or more embodiments of this specification also provide a device corresponding to the above method. Figure 3 As shown, Figure 3 A schematic diagram of the structure of an electronic device provided in an embodiment of this specification, the device includes:
[0057] at least one processor; and,
[0058] a memory communicatively connected to the at least one processor; wherein,
[0059] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to the first aspect.
[0060] Based on the same idea, an embodiment of this specification also provides a non-volatile computer storage medium corresponding to the above method, which stores computer executable instructions. When a computer reads the computer executable instructions in the storage medium, the instructions enable one or more processors to execute the method described in the first aspect.
[0061] This specification may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. This specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.
[0062] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, equipment, and non-volatile computer storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.
[0063] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0064] The above description is only one or more embodiments of this specification and is not intended to limit this specification. For those skilled in the art, one or more embodiments of this specification may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of one or more embodiments of this specification shall be included in the scope of the claims of this specification.
Claims
1. A standardized method for talent capability labeling, including: Get initial labels from natural language; Logically decomposing the initial tag to generate multiple sub-tags, and determining logical relationship operators between the multiple sub-tags; Determine the type of the subtag, perform standardization according to the type of the subtag, and generate a standardized subtag, wherein the type of the subtag includes a character type, a numeric type, or an enumeration type; The standardized sub-labels are recombined according to the logical relationship operator to generate a standardized label combination.
2. The method of claim 1, wherein: Determining a logical relationship operator between the multiple sub-tags includes: The logical relationship keywords included in the initial tag are determined, and the corresponding logical relationship operators are determined according to the logical relationship keywords.
3. The method of claim 1, wherein: When the type of the sub-tag is character type, generating a standardized sub-tag includes: Searching for a standardized subtag corresponding to the character through a standardized dictionary, and if it exists, determining the query result as the standardized subtag; Otherwise, the sub-tag is embedded into a vector and the related standardized sub-tag is queried in the knowledge graph vector library according to the vector.
4. The method of claim 1, wherein: When the type of the sub-tag is a numeric type, generating a standardized sub-tag includes: Determine the type of the numerical value, and convert the numerical value into a standard numerical value according to the type of the numerical value; The standard value is determined as a standardized sub-tag.
5. The method of claim 1, wherein: When the type of the sub-tag is an enumeration type, the generating of the standardized sub-tag includes: Analyzing the semantics of the subtags; A preset classification rule is called, and a standardized enumeration label corresponding to the semantics is determined according to the classification rule.
6. The method of claim 1, wherein: Before generating the standardized tag combination, the method further includes: Post-processing the standardized sub-tags; determining a test type for the standardized sub-tag; A compliance check rule is determined based on the check type of the standardized sub-tag and the integrity is checked.
7. The method of claim 6, wherein: Determining the type of the standardized sub-tag includes: Performing semantic analysis on the initial tag and its context to determine the inspection type of the initial tag; The inspection type of the initial tag is determined as the inspection type of the sub-tag.
8. The method of claim 1, wherein: The method further comprises: If a standardized label combination cannot be generated, the initial label is determined as an abnormal label and stored in the abnormal processing memory.
9. The method of claim 6, wherein: Before generating the standardized tag combination, the post-processing of the standardized sub-tags includes: The standardized sub-labels are reclassified as follows: Obtain the vector λ1 corresponding to the sub-tag, and obtain the remaining sub-tags λ2 in the first initial tag that have a logical relationship with the sub-tag to generate a first tag combination, the vector corresponding to the first tag combination is λ3, and determine the ratio of [λ1+λ2] / λ3, where [λ1+λ2] represents the concatenation of the two vectors λ1 and λ2; at the same time, obtain the remaining sub-tags λ4 in the second initial tag that have a logical relationship with the sub-tag to generate a second tag combination, the vector corresponding to the second tag combination is λ5, and determine the ratio of [λ1+λ4] / λ5; if Where K is a preset threshold constant, the sub-label is reclassified into the second initial label type, and the second label combination is generated with the remaining sub-label λ4.
10. An electronic device comprising: at least one processor; as well as, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 9.