SE2030053A1 - Integrated automation in matchmaking for business platform - Google Patents

Abstract

The service states the four main categories of personality assessment, logic / talent analysis, work experience and education in a mathematical matching algorithm. Through the demand for the main categories, a search engine finds information on profiles to achieve a matching result.

Description

Technical area A tool that auto generates a match via an algorithm for job-seeker to employerand vice versa without a third party. Specifically, for job seekers and employers toeasily visualize demand based on the structure and function ofthe tool.

Position When recruiting staff for a service, it is only about finding the right person for theright service, so that the service should not be very strenuous for the individual.This process is done by studying the individual's background and experience,thus comparing with the requirements set for each service. The individual whomatches most requirements is also the individual who usually receives theservice.

To match individuals to services, work experience and education are primarilyused, but also personality or individual intelligence. Because personality andintelligence are standardized according to scales, it is therefore easier to producea mathematical matching algorithm that links the individual's personalcharacteristics to demands placed on work.

On the other hand, work experience and training vary depending on the employeror training, for example the title ofa job, education or responsibility, division ofwork or training plan. This has therefore complicated the development of a fullyautomated matching algorithm and instead handwritten resumes are still used. Aresume is a great way to briefly convey experiences, CV, on the other hand, is a document that must be handled manually and thus itbecomes an individual-based match. An individual who handles job applicationswould probably with negligence choose a strong candidate.

Therefore, since a resume is a handwritten document by the candidate, falsifiedcontent of information may be likely. This may include overestimating theirabilities, overwriting their experience in writing, or wanting to convey untruth.The result of this action enables manipulation in the matching procedure.

Structure and content vary in a resume as they are individually based, and cantherefore contribute to misunderstanding or influence the impression whenselecting a candidate for services.

Problem solving The present invention solves the problems ofthe prior art by having the featuresset forth in the appended claims. The invention also solves the problem in thatmatching is an automated process which eliminates risks in an individual-basedsorting.

List of Figures Fig. 1 shows the overall functionality for matching between profile and job seekerrequirements in an advertisement based on the main categories' personality,logical thinking, education and experience.

Fig. 2 shows the ad's search for several profiles stored in the database to create amatching result based on the main categories personality, logical thinking,education and experience.

Fig. 3 shows the demand ofthe ad in relation to the characteristics of profileswhere the tree model distributes comparison values for the match.

Fig. 4 shows how the different levels of the level structure affect the final result inthe total match.

Fig. 5 shows the flow chart which is a visual representation ofthe informationflow during a match search as well as the path towards a match result.

Specific section The matching algorithm is mainly based on four main categories (1) personality,(2) logical thinking, (3) education and (4) experience. These four differ dependingon the individual's experience and thus affect (34) the matching result. The fourmain categories for matching are stored in each individual (5) profile. Theseprofiles are stored in a (15, 28) database.

The personality defines the personality traits of the individual and how he relatesto work tasks but also wishes made by the advertiser. These criteria are based ona personality test that defines an individual's character from a work perspective.The individual's intelligence is defined by a logical test, which is a result ofhowthe individual logically reason for different tasks. The training describesknowledge that the individual possesses either from a theoretical or practicalrelationship. Work experience describes the individual's work experience andperformance during work based on work categories. The experience is saved inthe profile based on previously performed as well as existing work assignments.

Data is stored in the same (15, 28) database for notjust information in theprofiles but also the set (11) requests in the ads. This means that the requirements placed in the advertisement within the categories (6) personality,(7) logical thinking, (8) education, (9) work experience, and thus a result isobtained on the hit security in the form of a (13) percentage figure via thematching algorithm. This is done through an automated matching procedure.

In order to increase choice, individuals have the opportunity in ad writing todefine which or which ofthe four main categories are important or imposerequirements to attract the right profiles. This means that the four maincategories are placed in four different (10, 33) "levels" where the selected as theimportant categories always end up at a high level figure, while the other lessimportant categories end up at lower level figures. By creating a "level" model inthe matching algorithm, another dimension is created in the selection options bythe advertiser.

The result based on the (33) levels is combined to give a total result on a match.By contrast, the impact on the results of the different (33) levels varies, where(23) "level 1" affects the majority of percentage points in total profit, (24) "level2" affects less than "level 1" but more than "level 3 ", (25)" level 3 "affects lessthan" level 2 "but more than" level 4 "and (26)" level n "affects at least onepercentage point of the total result.

In this breakdown of (33) "levels" (14), the total score ofpercentage points on thematch varies but more accurately based on the wishes selected in theadvertisement. This means that a 100% match hit in a category that falls below"level 1" gives the majority of the percentage in the total match, but never a fullhit (100% hit in the total match). To achieve 100% in total matching, theremaining categories must also receive a 100% match.

The matching algorithm provides the opportunity to reach the right individualsvia an auto generated matching procedure that compares the desires in the adwith the properties ofthe profiles. The comparison is done by the samecategories in the ad and profile. This means that (13) personality category iscompared with each other, category for (13a) logical thinking is compared witheach other, (13b) education category is compared with each other and (13c) workexperience category is compared with each other.

The search engine is the core ofthe matching algorithm where it (16, 17) locatesprofiles stored in the (18, 28) database and compares the data which is then sentto the visual result mechanism. The engine uses a search algorithm that matchesinput data with existing data belonging to a user's properties in the form ofsubcategories. When an entry is made, a search is initiated in the (28) databasewhere the algorithm recognizes all subcategories in the cloud and returns a valuefor different combinations ofsubcategories that a user may conceivably have inthe four (22) main categories. The different combinations ofthe subcategoriesare fixed in the (20, 32) tree model via its interconnections / branches for each individual (19) main category. This prepares and visualizes a (34) match resultwhen a pregnant user visually encounters an advertisement.

Figure Description Difficulty factor - Categorizes the properties into different (3 3) levels based onthe time and skills required to achieve them. Similarities and differences in workroles. Level value - Describes the level (29) of the input values subcategories /properties are categorized. (27D) Personality Value - Describes the aggregatevalue that results after a personality test. (2 7A) Educational Value - Describes theindividual or composite value of one or more programs. (27C) logic value -Describes the compiled value that results after a logical test. (2 7B) experiencevalue - Describes the individual or aggregate value that results after the numberofyears the experience receives. (29) Input values (27D) personality value, (27A) Educational value, (27C) logicalvalue and (27B) experience value is processed and categorized into a (33) levellist. The (33) level list receives all ofthe (29) input values subcategories dividedinto different (3 3) levels based on a difficulty factor in meeting theserequirements. The difficulty factor is based on the time and skills required by theproperties. Each (29) input value contains several subcategories that describe theuser's personal characteristics. These categories are what determine the sum ofthe (29) input value. When an entry is made by a searching user, the informationis assigned to a (28) database where a matching algorithm calculates thesimilarity between the searching user's input and the personal user ofthepregnant user. The matching value is shown as a percentage where one hundredpercent is the maximum value and zero percent is the minimum value. Theprocess constitutes (29) the subcategories ofthe input values as the base. Thesubcategories are a compilation ofproperties divided into (33) levels. (33) levelsdetermine the final value in the match where the properties, depending on whichlevel they end up with. Levels 1-4 describe the properties ofthe properties in thealgorithm where an exponential percentage value is deducted the lower or higherlevel the property is in.

All properties of the algorithm are assigned in an algorithm tree where theinterconnection ofthe properties is obtained via similarities between them. Theinterconnection determines the percentage value in the similarity based on thenumber of steps in the algorithm tree the properties are positioned apart. Thealgorithm tree links the properties of a systematic network and describes theproperties interconnection in percentage value that is supplemented into thefinal value, which is the match value.

Each individual subcategory is implemented in a (32) tree model where allsubcategories share a mathematical link with each other. This mathematical linkis based on the subcategory family level between two or more subcategories.During the procedure, a subcategory is sought from the applicant user, hence thepregnant user does not receive the subcategory among its properties but obtainsa subcategory closely related to the requested and thus creates a useful (34)matching result. Should the applicant and pregnant user receive the samesubcategory, this results in a 100 percent match for that particular search. Whena searched subcategory does not find a corresponding subcategory via the (31)search engine, an intermediate calculation is used for the searched subcategoryand the subcategories that the current user obtains. In this context, a calculationis made using the number ofbranches that differ against the subcategory appliedfor and thus constitutes a percentage extract for each branch that differs betweenthe searched and the subcategory.

The matching search procedure thus undergoes a (33) level classification checkfor all subcategories involved, thereby sorting them out by an add-on calculationbased on the subcategory's placement in the (33) level classification. The highest-ranking subcategories are placed in Level 1, which identifies them as the mostdifficult to obtain characteristics. The (33) level classification counts down from a(33) level list for all subcategories and the supplemental calculation for thematching algorithm utilizes the level placement of the subcategory and thusinitiates a deduction ofthe percentage result in the search. Should subcategory bein Level 1, no deductions are initiated, but all subcategories from Level 2 anddown will go through this procedure. process Description A user selects input for (27D) personality value, (27A) Educational value, (27C)logical value and (27B) experience value according to the requirements the userrequests. The user can apply an absolute requirement which means that thematching user requested must obtain these requirements for the property. Anabsolute requirement can only be applied to (29) input values for experience andtraining. Once the employer has selected the requirements specifications for theproperties and approved the application, an advertisement is created containingthe (29) input values. These (30) values are automatically categorized under the(33) level list and then create a (29) input value as a whole that is linked to thesearching users.

The employer user can only see the values obtained by the application in the formof property description and not a numeric value. For the applicant user when theadvertisement is visualized on search, a matching value is created which thealgorithm automatically calculates via the connection between the employer user's requirements ofproperties and the applicant user's personal propertiesobtained during personal data.

Claims (8)

Claim

1. Designation with intent to characterize functionality of the four maincategories ofpersonality assessment, logic /talent analysis, work experience andtraining in mathematical matching algorithm, characterized by personalityassessment in matching purpose, constitutes user's personality matching inaccordance with the algorithm's performance test, encoded in algorithm theuser's problem-solving properties and reasoning encoded in algorithm after agiven logical test, experience in matching purpose constitutes the user'sexperience encoded in algorithm both from previous and existing work andtraining in matching purpose constitutes the user's education history encoded inalgorithm by number ofissues issued and ongoing.

2. Procedure according to claim 1, characterized in that the purpose ofthe fourmain categories in advertising grants via users constitutes the advertiser'sdemand as a starting point for (34) matching results, whereby the correspondingstarting point for applicant users emanates from the applicant's personalcharacteristics in the main categories.

3. Procedure according to claims 1 - 2, characterized in that the main categories'collection data in the starting points obtain information networks in the form ofsubcategories thereof, a (31) search engine for the location and identification ofsought subcategories and visualization of (34) matching results in percentagepoints between two parties.

4. Procedure according to claims 1 - 3, characterized in that the subcategories ofthe main categories are re-examined for (33) level classification and assigned in agrading scale based on accessibility and degree of difficulty to achieve ownedproperty within said subcategory.

5. Procedure according to claim 3, characterized in that the individual percentagematching result between the applicant and the advertiser user's properties in themain categories, describes the similarity in the properties between the twoparties thereof, if two properties from different parties are identical, this resultsin the highest possible match.

6. Procedure according to claims 3 - 5, characterized in that the subcategories ofthe main categories are linked in a (32) tree model where the branches ofthemodel explain the relationship between all subcategories and are decisive in a(34) matching result based on the relation ofthe desired subcategory to the otherthe party's holding subcategories.

7. Process according to claim 6, characterized in that the (32) tree model'sfunctionality describes the algorithmic search of subcategories during a matchingprocess, where the (31) search engine finds the requested subcategory in the(32) tree model, locates the other party's subcategories in the (32) tree model.and calculates relationship in percentage unit based on number oflinks betweensubcategories.

8. Procedure according to claims 4 and 6, characterized in that the (33) levelclassification of subcategories is taken into account (32) the functionality of thetree model where an additional calculation is established in the algorithm withthe intention of describing the subcategories weight in the matching process andcalculates the actual (34) matching result.