TW201003434A - Manpower data mining association rule and the system thereof - Google Patents

Abstract

The invention relates to a manpower data mining correlation rule and the system thereof, characterized by exploring the status between manpower-hunting conditions of the network bank. By virtue of using a data mining method, the relation between manpower-hunting conditions is analyzed to further understand the association between enterprise's demand for manpower and job-hunting conditions, which is especially applied and connected to the exploration in technical aspects of association rules, wherein the technique of association rules is used to find out mutual association between some commodity items in the database. This information is utilized properly to then effectively help the user to discover implicit potential associated knowledge and provide a real-time reference for two-party mutual matching between job-hunting and manpower-hunting sides during the process of employment. Furthermore, the purpose of the use for manpower-hunting analysis generalization and conditional association index can be attained.

Description

201003434 IX. Description of the Invention: [Technical Field] The present invention provides a human resource database mining association rule and a system thereof, wherein the key to the association law exploration is to find out possible variable items in the transaction, and the architecture of the present invention mainly includes four Item processing variables; data collection, data exploration, correlation analysis and analysis conclusions, by sampling sample data, according to the method of selecting association rules, analyzing the information to be mined from different viewpoints, through understanding the related trend The target _ demand, in order to determine whether it meets the characteristics of the definition of the reduction of the conditions between the conditions ' 俾 integrated the overall demand environment preferences, the real-time response to the conditions associated with the job seeker as a reference for the application purpose. Prior art] The job-seeking _ is only to announce the content of the conditions for the recruitment of the enterprise (please refer to the figure--), in order to improve the application of the conditions of such enterprises, the ship can be relied upon. Conditions and skills to find work that is consistent or similar, or to allow applicants to know which primary and secondary conditions should be met for a job. 'Enable the applicant to strengthen the insufficient part in advance, so as to increase The application rate is 2. It will enable applicants to find out more quickly what they can do to find the relevant talents they need. It is also possible to let the students learn to know what they want to do in the future. Gam == 201003434 It is possible to develop the training for the talents needed for the difficult reading in the vicinity: the employment opportunities and work ability of the strong students after graduation. The head of the personnel and the employer of the employer also said that they are all _ single _ special Wei Lisi _, market supply and demand is

There is a gap in the current view of the company as a recommendation and reference for the conditions of the approval and the direction of the evaluation. X data: = The method of depreciation is mostly statistical analysis. So far, the increase in the number of breaks, if only through the simple statistical method has not met the need to go back to the application of the method is to pre-empt the problem conclusion _ ^ ^ Correct, its system verification guides her rif_-(5) Movement's use of data mining technology, through automatic, semi-self-legal logic == This is the job of the above-mentioned statistical party, but Before the emergence of the solution, data mining still needs more complicated calculations. The difference between the two cannot be guessed, as shown in the following table: 1.-----__ Table 1 ____ analysis factors And project data exploration method. :-----j method----- must specify the definition of the data attribute must be _^ determine the problem target must be ^7\ must provide analysis algorithm ----- statistical analysis method , artificial intelligence 'decision tree, neural network ------ statistical analysis method mode establishment - provide a variety of modes, can be determined within the appropriate 53⁄4 art analysts 201003434 --~~--, ~~ one, ~ ~~~~~~_ The person wants sex, the mode is Establish correlation variables ---- can find the correlation between multiple variables Γ - 只能 can only check the impact of one variable on the result at a time can anticipate the analysis results ------ can be executed ~~__ no loop, The process of continuous revision - problem-oriented, the same problem usually only needs to be summed up. 'The data mining is similar to the same way to find the guidance ((10) warm job). Without the assumption of the target result, it is impossible to verify the relative _ of the analysis. It is not applicable to the technique of statistical verification. How to effectively use the statistical method of Data Mining is exactly what the present invention wants to & The technical characteristics of the improvement. According to the US Patent No. 6,735,571 "A System and Method for Providing an Estimated Salary", the invention of the patent statement reveals that "this secret includes - communication network interface, - storage and processor - communication network interface for consumption Into the communication network. The faulty device is used to store the rhyme data. The H-wire-to-orbit network interface=storage device is processed, and the regression factor of the pay-per-payment is confirmed by the linear unit, multivariate or curve multiple regression method. Establish a regression formula to calculate the salary of each job."; The theory of the invention is discussed. The towel is directed at it. _ Rong Lin: "Get salary data, check the standardized salary structure and adjustment, and the impact factors include seniority, industry. , public 51, market salary reference, etc. Wei, through the future estimated salary and the actual salary ratio obtained by the state towel, with the regression reference factor as the adjustment factor 'when the pre-settlement and the actual fine are secret Acceptance scope, the return of 7 201003434 The correctness of the returning party's level of the future can be supplemented by other factors' Applicable level of funding. ". However, 'the prior art of the other is not able to dig into the structure of the salary structure of the acid, and the law is adjusted according to the curve of the curve, and under the interference of the dynamic environment changes, the job is adjusted, promoted, transferred. In order to be proportional to the estimated salary, in view of the fact that the case is based on the theory of the database (Da1; abase her resistance w) theory, through the application of inventions to deal with the simplification of the day, derived - more objective Accurate data reference 'Based on this county is the technical characteristics of the case of the case of the case. [Invention content] The situation between our H network human bank's seeking conditions in the month of the month, and through data exploration (touch The technology, analyze the relationship between the conditions of seeking talents, understand the relationship between the talents of the enterprise, the ___, _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ Block one data selection 1〇, which refers to the data pre-processing position; - Data purification 20 'Re- (four) material county, some records in the data are blank spaces, some exaggerated false or wrong, must be in the process of analysis Will this Some of the noise information is excluded, so as to avoid the inaccurate results of the analysis; a rich 3Q, a variety of wealth _ solution sharp set, lion generated Guan 8 201003434 joint law is in line with the target ugly demand; Code 4〇, in the case of phase and complete equipment, and set the relevant numerical parameters; shell material mining 5G, combined with: #继Μ_) technology, analysis between the conditions of the _, to understand the enterprise The relationship between the job-seeking conditions of the talents, expecting to find useful and important potential relationships; and reporting 4 estimates of the conditional characteristics analyzed according to the association rules, integrating the entire environment of seeking talents and company preferences to provide the conditions for the applicants to be associated with Reference. [Cross-application method] ^Zhangshe wants to describe - manpower (four) library mining _ rules and its system, in order to make: Chapeng easy to ride to understand the other special features and advantages of the invention and its effects (four) more For the sake of manifestation, return to the ship to ship _, the detailed description is as follows: In the data excavation, _ Fu rule (AssQeiatiQn plus (five) is one of the most commonly used = technical methods, in the #material mining algorithm, is A method that can produce a large 罝^J. The algorithm of Apri〇ri's Lailian rule (Agrawai, Imielinski, and Swami) was first proposed by Min Ji! and other scholars. Its main function is to find out the data (4) __ At the same time, the general law (Cai (4), i-ki, ands job i, 1993) is based on the degree of support (s_〇rt), reliability (or confidence, 'Confidence'). For example, According to the product information that the consumer in the store buys every time the product is bought by the father, the age can be mined. _ The law of the joint: If the customer of the job considers the toner g, the report paper will also be considered at the same time. 201003434 2 Question ^ is as follows: Let 1 be the collection of item items sold in the store in the parent database, each transaction contains a transaction number and a set of items that are made by the group of items. Call it the "project set" Η-. For a project set, if all the items in X are included in the transaction τ, then we call the transaction support set X. The number of support items in a project set X is defined as , and supports the transaction of the project set X. The support degree of the project set is, and the number of transactions supporting the project set accounts for the total number of the masters. proportion.

Continued the form of the association rule Π 4 γ [support, confidence level Guangqi and Υ representative project set, we call it "conditional sentence" γ is "conclusion w,. The association law X 4 γ support is defined as the project #χυγ的支持度,和^联贝j X — γ confidence level (c〇nfid_) is the ratio of the number of transactions that satisfy the conditional sentence and the conclusion of the parent of the p. Confidence level ==xuy + X support degree 2 "support" and "confidence level" mentioned in the t, respectively, are the two criteria for the attack law, the two standards are usually evaluated as _ The standard of establishment; 'the threshold of support and confidence level is too high to be easy to produce rules, easy to cause omissions. If the threshold is too low, there will be a lot of confusion: a reliable platform's support and confidence. The set value of the transfer needs to rely on the experience of the analyst to achieve accuracy and practicality. The association rule must satisfy the previously set parameter values: minimum support (minimum support) and minimum confidence level (five) (1) difficult c〇nmence). 201003434 The product of minimum support and the total number of transactions in the database is the minimum support count. Assume that the minimum support and the minimum letter d are assigned Q, 2 and Q. 5; the number of support for the association rule {1,3} {5} is 2 ', so the support is 〇. 2, and the project set is ,1. The support level of 3丨 is 〇.3, so the correlation rule U, 3}—the confidence level of {5} is 〇. 2/0. 3=0. 67. The association rule _ questions can be subdivided into two sub-questions: First find! ϋ All items with support greater than or equal to the minimum support, called "large project set, (largeitemset); then from the big 31 project A correlation rule that produces a level of confidence greater than or equal to the minimum level of confidence. Suppose Z is a large project set, and the financial form is χ-γ, which satisfies χυγ=ζ,χηγ=

Health cuts to the least confidence turn _ Lianfa _ should be produced ί will == Ming project 编 的 的 的 的 产 ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' ' Between the eight conditions to understand the relationship between the company's money and people seeking to find the relationship between the strips, i 4 conditions __, expect to find useful weight 2 potential _, secret money (four) is _ including the following six steps: Information 撰 Λ Λ , , , , , 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 、 The application of a related law technology, by 11 201003434 201003434, to effectively achieve the analysis of the disclosure of the following: The purpose of the use of the demand, the use of the law of various related laws and techniques (Muitiievei Associati〇n Ruie ing) The class conducts cross-level exploration, and gives different levels of different J, support degree of flexibility. This association rule discovered by the use of the similar class is called "Multi-level association rule, Onultilevel _ciation rules" In the shirt should In the use of non-age class to explore the association rules is very useful. For example: Fabe]. If you buy a pc host, you will also buy a screen. Rule 2: 70% of customers who buy IBMpc host, At the same time, I will buy a 仏 沁 screen. The second rule is expressed by a lower concept level (1〇wer c〇ncept ievei), but it can provide more detailed information than the law. Although it is 'lower' 〇 The concept hierarchy represented by 概念_) can be higher than “higher,” (highe0 concept c level presents more information, but may result in “lower support for item 7 at lower cost” If you want to explore association rules at a lower conceptual level, the minimum support level must be reduced. For example, HP Business Computer-> ViewSonic LCD screen (support = 〇〇1) but this approach may The actual benefits of the law will be greatly reduced. The association law generated at the higher conceptual level, although having a higher support voice, may be an obvious result that can be predicted by experience. — 12 20 1003434 For example, people know that the screen (support degree = 〇. 95), the actual benefit of this obvious result is not high. The basic spirit of multi-level association law exploration; using the top-down approach = first calculation At the level 丨 (level 0 large project set, then calculate the large project set in layer 2 2 (levei-2), and so on, until there is no type of project set. In each level You can use the Apriori algorithm to generate large project sets. ^ When you perform the money association method, you can make the minimum support level different for the level. You can also add the following two conditions to the condition limit: (1) At the level / Project, only when the project represented by the parent node of the 4 level yi is a large project, the χ need to be considered. example! : er f, in the class 1 "desktop computer, a large project" is the use of computers and "commercial computers" only need to be widely tested. Conversely, if, computer, not a large project, then "home belt Peng's heart table does not have to be considered. Use (4) and "commercial computer" (2), consider the multi-level of parent node support, only when it is in the level η ^ survey / consider the level of the outline, especially Lang points is a large 1-item ' when the project is considered to be considered, otherwise it will be ignored and will be provided in addition to the different conditions. 13 201003434 (b), consider the parent node left-project set support Hierarchical exploration: When considering the left-item set I of the level y, only when it is a large project set of the parent node corresponding to the level /-1, j needs to be considered, otherwise it will be ignored. The following is a brief description of the processing steps of the multi-level association rule exploration algorithm: Step 1: for (Lu 1 ; Z[/, 1] #0 and / maximum level; /f+) do begin /* starting from level 1 Generate a large set of items at each level V Step 2: if /=1 then { Step 3 : 1] = Large"tem_gen(7[l],/) ; /* Find the large 1-item set in level 1 from transaction database 7[1]*/ Step 4: 712] = Filtered_table(7[ l],Z[l,l]); /*Use Z[l, 1] to filter 7[1]*/ Step 5: } Step 6: else Z[y·,1] = Large—item_gen (7[2], _/); Step 7 ·· for (female=left++) do begin /* produces a hierarchy/larger than the project set*/ Step 8: Or = Candidate_gen(Z[/, left-1 ]); Step 9: For each transaction in 7[2] ί Step 10: For each candidate item set c in α, if ί contains c, increase the number of support for c by 1; Step 11: Z [y;left]=A set of candidate left-item sets satisfying the minimum supported number minsup[/j in fit; Step 12: end Step 13: return ZZ[/]=All large items in the hierarchy/ The collection step 14: end 14 201003434 According to the above processing steps, an embodiment proposes a practical application: Related definitions: 7 is the original transaction database. 7Π] is a collection of items that are converted from the transaction in the transaction [/, phantom representation in the level y• of the big item using the project encoding method. [/] is the collection of the item set in the layered illusion. Minsup[/] represents a collection of large-scale project sets of the hierarchy. (7) Suppose there is a number of _ article number 3: ^. , Lai": counting, ϊ: table knowing 13⁄4 device. Representative printer", number "丨 represents representative 5" generation 1600 1601 1602 1603 1604 1605 1606 1607 1608 portable electricity month g_ brand

IBM

COMPAQ Home Ray BiL<

Tablet Linghuang --- Earth pole Table - τ_·_ entanglement transaction data employment transaction number Item number 1 {1600,2301} 2 {1602, 3457,4563} 3 {1601,2305, 4563} 4 {1606,2302}

ASUS

HP

IBM

ACER

IBM

ACER

TOSHIBA 15 201003434 5 {1605, 2302, 46231 6 {1608, 2306} 8 {1606, 2302,45631 9 13001,54531 10 {5455} The class is re-edited,

For example, 'the first transaction in Table 2 contains the item with the item number "16〇0". According to the personal computer classification description in Table 3, it corresponds to the IBM household brain's and then according to the classification level of the personal computer. The corresponding code is "1ΠΓ° csf in the personal computer classification class, which can be represented by the following coding method: . Under the representative, the lower level represents the l computer", the second number i more spoons | The type of computer,,, the third arsenic digital branch "commercial computer", the fourth to the pM business computer was expressed as "1122", which in the first level of the "personal, the first branch of 2" first Level 2 of the second "commercial computer" in the second branch of level 4 "IBM computer" after the transaction database 7ί 11 transaction number coded item set 1 {1111, 2111} 2 Ί {1113, 3214,41111 3 ” {1112, 2211,4111} 4 {1212,21121 5 {1211,2112,42121 6 {1221,2231,52141 7 {1222, 2223} 8 {1212,2112,41111 9 {3112,5211} 10 {5213} 16 201003434 In step 3 ' Large_item_gen (7[1], /) program's purpose From the transaction database mustard i], find the large-scale project set in the hierarchy/, ie Z[乂丨]. ◎ When considering level 1, use Large_item_gen(T[l], 1) to generate large-scale project Set Ζ [1, Π. ^ @ 层级/(y>1), use Large" tem-gen(7t2]," to generate a large set of items Z[/, 1]. @有在z[/—丨, 1] the project can be considered in the 4·7·, 1] classification class as an example, in the level 2 of the "desktop computer, the second 3 'two gas: table ten type computer" is not a large project The set of 'fit' computers ("111*") and "commercial computers" (112*) are not considered. In step 4, the purpose of the Filtered-program is to use Z[U] to check 7[1] Each transaction in the transaction and the following actions are performed: Θ 彳 ί ί any support items less than the minimum support health. ◎ If ί does not contain any large projects, then remove the force from the fairy. mtered The purpose is to reduce the size of the database, and the results of its execution are stored in the database. 2) The above-mentioned 'f first from the layer: the magical open bribe produces a large set of items at all levels. The minimum number of support for the layer-person 1 is 4. After the easy database, the large 1-item set of the layered illusion ' ^ 目 , , ί 4 *** } appears in the transaction number "2", "3", 17 201003434 • Supported in 2) The number of supports deleted during the filtering process is less than 4 'So this item • The transaction number is 9” and the items contained in -B-S a are not large items 1 if ff The packet filtering process was deleted. And the two fathers will also be in the second. The quantitative and succinct law exploration (Qu magnetic also (four) Α·^ί〇η test

Mmmg), considering both the item and the quantity of purchase, the set of items that will create the minimum support becomes less, and may not even produce any association rule that satisfies the minimum holding degree. For example: the relationship between the item and the quantity of purchase It has an important impact on the decision of the marketing strategy. ◎ For example, 40% of customers will purchase three packs of report paper if they purchase a toner cartridge.” We call this rule “Quantitative association rule” Quantitative association rule helps decision makers Develop a more effective marketing strategy. ◎ For example, when considering the promotion strategy of toner cartridges, you can propose a lock-up scheme similar to “Buy a toner to send two packs of report paper” based on the information of the survey results. Consider adding a number of intervals to the Intervals to increase the support of each project in its own interval to discover more potentially useful association rules. Its related 疋 meanings include · 18 201003434 (1), assuming that each transaction in the database is composed of a transaction number and a set of α-items (<3Litem), the form of the ^_ item is . It represents the item y and the number of bets bought. (2) The set of 1 set of items is called the t item set (g_itemset) ° (3), and the support of a q_ item is defined as 'the ratio of the total number of transactions to the total number of transactions.'... Project χ - the purpose of 'has a phase _ purpose u may have a different number

The support level may not be f small 'If most of the I project support can quantify the number of smuggling excavations, we will be all the possible "quantity" of the mother-item project in each project The finer step. Knife Che (4), borrowed and owed 1___ lift - the actual description is as follows: Now the table five t number is 2 'Based on τ table four to find all large project sets, presented \ - ----~~~- < 7 item ------- Project's land information~~ _. TS ^-~- SP <Α, 1> 3 <Α, 2> —----- —---- JM0J5} 3 <Β,1><Β,2> -—- 4 il〇iiU2,13,14} ——— ——— — 5 <Β, 3> ----- iMilL 3 <C ,1> -——J JiAi§,7,8,13i 7 19 201003434 <C, 2> {4,10,11,14,15} 5 <D, 1> {2,7} 2 < E, 1> {7} 1 <F, 1> {1,7,10,11} 4 <F, 2> {3, 9,14} 3 <G, 1> {2, 3, 4 , 12} 4 <G, 2> {1,8,13,14,15} 5 Table 5: Large 1-L project set buckwheat project set TS SP {<A,1>} {5,12, 14} 3 {<A,2>} {4,10,15} 3 «Β, 1>} {1,4, 5,8} 4 {<B,2>} {10,11,12, 13,14} 5 {<B,3>} {6, 7,9} 3 {<C,1>} {1,2, 3, 6, 7, 8,13} 7 {<C, 2>} {4,10,11,14,15} 5 {<D,1>} {2,7} 2 {<E, 1>} {7} 1 {<F, 1>} { 1,7,10,11} 4 20 201003434 (<F, 2» { 3, 9,14} 3 I i<G,1>丨" {2, 3, 4,12} 4 i<G,2>_| ” {1,8,13,14,15} rd in large projects Sewing the test, making the job β, and whether it is a large h project* = Α generates information about the evasion 1 project set project. Wood, Table 6 shows all large 2-β ΚΜ^<01>} ΚΜχ<22>}{^2>^2^{<Β,2>,<F,1>} KCJ^Dj^j. {<C, 1>, <F, 1>} A KCJ^ <GJ>} {<C, 1>, <G, 2>}

1〇^&gt;^2&gt;} ΚΑ^χ&lt;〇2&gt;} 201003434 {&lt;C,2&gt;,&lt;F, 1&gt;} {10,11} 2 {&lt;C, 2&gt;, &lt;G , 2&gt;} {14,15} 2 In the large 3-g_ project set generation phase, use the information in Table 6 to generate the candidate t item set &lt;3, and determine whether it is a large 3-&lt;7_ item set. Table 7 shows all the large 3-(?_ project sets related information, Table 7: Large 3-ζ?_ project set large 3U staff set TS SP {&lt;B, 2&gt;, &lt;C, 2&gt;, &lt;;F,1&gt;} {10,11} 2 {&lt;B, 1&gt;, &lt;C, 1&gt;, &lt;G, 2&gt;} {1,8} 2 Since 64=0, Ζ4=0, so End the generation process of the large project set. Continued above, assuming a minimum confidence level of 0.65, the quantitative association rule that can be generated using the corresponding information in Table 7 is as follows: {&lt;Α,1&gt;} ◊ {&lt;Β,2&gt ;} Confidence level = 2/3=0. 67 {&lt;Α, [2.. 3]&gt;} ◊ {&lt;C,[3.. 4]&gt;}Confidence level=3/3=1 { &lt;B,[3. .4]〉} ◊ {C,[1..2]} Confidence level = 2/3=0. 67 Confidence level = 2/2=1 {&lt;D,1&gt;丨◊ {&lt;C, [1..2]〉} {<B, 2&gt;,&lt;C, [3.. 4]&gt;} ◊ {&lt;F,1&gt;}Confidence level = 2/3=0 . 67 {&lt;B,2&gt;,&lt;F,1&gt;丨◊ {&lt;C,[3.. 4]&gt;丨Confidence level=2/2=1 {&lt;C,[3.. 4 ]&gt;,&lt;F, 1&gt;} ◊ {&lt;B,2&gt;}Confidence level=2/2=1 {&lt;B,1&gt;,&lt;C,[1. · 2]&gt;丨◊ {&lt;G,[3.. 5]&gt;}Confidence level=2/2=1 22 201003434 {&lt;B'1&gt;,&lt;G,[3. .5]〉} ◊ {&lt;C,[1..2]&gt;}Confidence level=2/2=1 {&lt;C,[1.. 2]&gt;, &lt;G,[3. · 5]&gt;} ◊ {&lt;β,;[〉丨 confidence level=2/3=〇. 67, the application of the above related correlation law exploration, and finally the statistics of association analysis, To obtain the conclusion of the target data mining, the correlation analysis process can at least reach the effective information profile: (1) The association rule that satisfies the minimum support level and the minimum confidence level does not necessarily guarantee that the specific useful information can be provided. In the database containing 10,000 transaction records, there are 6 transactions including slaves, 7500 transactions including report paper, and a transaction containing both toner and report paper. Assuming a minimum support of 3〇% The minimum confidence level is 60%. The following correlation method will be misleading: Toner Report Paper [Support = 40%, Confidence Level = 67%] ◎ The probability of purchasing report paper is 75%, which is greater than the confidence level of 67%. However, buying toner cartridges reduces the possibility of purchasing report paper. (2) If = eight j) eight fishing, it means that the occurrence of project set j is independent of the occurrence of the project set, otherwise it means that the project set J and the project set are related (dependent and correlated). ◎ The degree of correlation between project set 4 and project set (correlati〇n) is calculated as follows:

CorrA, B = PCAuB) / ΚΑ) Ι\β) If correlation &lt; 1, it means that j and the square are negatively correlated, that is, the appearance of J will cause the probability of 10,000 to decrease. If correlation &gt; !, it means that j and the square are positively correlated (p〇sitively correlated), that is, the appearance of J will cause the probability of 10,000 to increase. 23 201003434 If correlation = 1, it means that j and 厶 are irrelevant. When the degree of correlation is greater than 1, it means that the effect of using the rule will be more significant. However, when the correlation is less than 1, it means that the effect of using the law may be poor. In summary, the present invention seeks a job-seeking platform based on the human banking network, relies on the efficiency of the network speed, cooperates with the government to rescue lost materials, assists job seekers to enter a good enterprise and helps the talented enterprise to be the fastest. To find the talents needed to meet the needs of manpower in all walks of life, and to let the unemployed or unemployed people find the work they want as quickly as possible. The data mining practical application is based on the steps of data mining. The actual calculations are based on the analysis of the conditional association. First of all, we define the problems and needs. In the existing job-seeking channels, the network manpower bank mainly uses the population of the Internet as the main county, and the enterprises with the wealth and the job seekers must have the The ability may be different from the traditional companies and talents in the traditional job search pipeline. Therefore, the purpose of data mining is to analyze the conditions of the talents in the human bank database. 'Analyzing the conditions that often occur or the conditions of ff-lifting, and matching the results with the school_planning, _ bank is turning the county Seeking talent. The following application data is provided by the human resources database. The data is for informational design, (10) design, and the requirements for the exploration of the valve. A slightly needed field: the human bank database, and her multiple barriers in the stomach are not applicable or need to be processed before they can be used. The block used in the present invention has 0/1 201003434 A_CodeNameA &gt; B_CodeNameA ^ C_CodeNameA ^ D_CodeNameA &gt; _C〇deNameA, and combines the five blocks into one and gives the number. (2) Some of the records in the data of the miscellaneous materials are blank, some are exaggerated or wrong. In the process of analysis, these noise data must be excluded to avoid inaccurate analysis. (II) Association Rules Mining and Rule Analysis 'In the case of relevant data and complete equipment, according to the data to be mined, the relevant values are set. The computer expertise of the job conditions required for this job is the execution condition. After screening, The total data of 8862 is SQL2〇〇5. The SQL2005 correlation rule program is used to process the filtered data. The results are as follows in the definition of the project set shown in the third figure: Minimum support: This parameter That is, the minimum support level in the association rule. Objects that support values below this parameter will be filtered. Taking MySQL as an example, MySQL has a total of 981 times in all the data. Minimum item size limit: The number of items in the item set below this value will be exceeded. Show full name: Check this box to display the full name (including the data table name) of the project set content. Filter the item set: Press ENTER after entering the keyword in the box, then _ select the item set containing this keyword. 25 201003434 Display: You can switch the display attribute name and value. If you feel that there is a lot of "xx=existing" in the face, you can switch to "Show only attribute name". Maximum number of data columns: Displays the number of items that the viewer can display. The definition of the relevant rules of the project set shown in the fourth figure: Dare to take the opportunity. This parameter is the minimum confidence level in the association rule, and the rule that the ancient heart level is lower than this value will be transitioned. The minimum importance: the high probability does not necessarily mean a meaningful rule. For example, “Is there 80% of B with A.” Does this rule have any meaning? If you randomly find a company to find a condition with a chance of B is only 20%, then this is a meaningful rule (find a group that is more prone to ambiguity), if the probability of a stochastic enterprise is as high as 90%, then this time The rules are not meaningful. We must compare the probability of occurrence of sputum under the ambiguous conditions, because this ratio may be quite disparate. Therefore, by taking the logarithm to pay the importance index, the formula is as follows: Importance Μ => 扪 = The probability of embarrassment under bribery conditions] 'In the absence of a gallbladder rate 13⁄43⁄43⁄4 26 201003434 According to the formula 'When the probability of A occurring under B conditions is higher than the probability of A occurring under the condition of B,' then the pair It will then be greater than zero, and the larger the metric, the more significant this rule is. Conversely, if it is less than zero, it means that A has an inhibitory effect on the occurrence of B. In SQL 2005, the "support" data commonly used by association rules is not directly calculated. You must switch to the "Project Set" page, as shown in the third figure: (The second column data MySQL and MSSQL occur simultaneously It is 664, and the total number of data is 8862. The first item of the "rule" option (MSSQL_&gt; MySQL fourth picture) can be interpreted as: "If you need MSSQL, you also need MySQL support =664/8862=0 · 074926=7. 4926%, reliability = 40.6%, importance=0. 965" and related to the related dependency network definition of the project set shown in the fifth figure: The dependency network is for The user understands the graphical view of the correlation between the variables. Each arrow link represents the predicted relationship (the starting point is the input variable, the arrow 碥 is the output variable), and the predictive association between the variables can be understood based on the strength of each arrow link. The intensity of the sex. On the left side of the fifth picture, we can use the scale switch to view the more relevant links. You can also click through the mouse to understand the dependencies between products through the graphic color. The purpose of this embodiment is to carry out the four processes of the town: data collection - sample data provided by the human bank database. 27 201003434 2_ data exploration - selection of association method. The way is to explore the conditions of recruitment Between the exchanges and the news, I understand the need for the target problem. The original definition is to determine the characteristics of the conditions, and the post-experimental association rules and job-seeking conditions are related. Whether the conditions are the ones that the company will first consider when actually seeking a job is whether it is selected by the general company; analyze the law of association from the conditions, position, company name, applicant information, etc. If the law is unreasonable , incorrect or non-conformity, then need to go back to the previous stage and then re-run. 4. Analysis and conclusions - based on the conditional characteristics analyzed by the association law, comprehensively the entire talent and the company's preferences 'provide the application conditions associated with the job seeker The above detailed description is directed to specific embodiments of the invention, which are intended to limit the invention Lilai, the equivalent implementation or modification of the present invention should be included in the patent scope of this case. In order to make the invention more progressive and practical, a comparative analysis is used as follows. : Lack of usage 1. The user's career goal is poor, and the workplace change response is too slow. 2. For the current job search and just the scholar's adjustment guide for the future ideal career planning is not enough. Advantages of the present invention 28 201003434 With objective and accurate pre-l, the user is provided with the superiority of the measurement and evaluation application according to the individual's future career planning goals.

Combining the data mining theory with the analysis of the data, it is suitable for the reference of the trend of the world. The user guides the student's job description. The invention has achieved the desired effect under the previous technical structure, and it is not familiar with the face and the face, and the application is not before the application. Public, its progressive, practical, has been in line with the application requirements of the invention patent '爰 apply in accordance with the law, I ask you to approve the invention of the invention ^ application" to encourage invention, to the sense of virtue. [Simple description of the schema] Simple description of the schema | Schematic diagram of the content of the announcement of the job-seeking website. f is a schematic diagram of the implementation process of the technology of the present invention. The definition of related content of the project set of the embodiment of the present invention.相关 The definition of the relevant rules of the project set of the embodiment of the present invention. The fifth figure is the definition of the relevant dependency network of the project set of the embodiment of the present invention. [Description of main component symbols] 10 data selection 20 data purification enrichment 40 series stone horse 50 data mining report evaluation 29

Claims (1)

201003434 X. Applying for a patent garden: Bu-type manpower data _ _ Ma Maqi system, the process steps are: - data selection 'the system refers to the pre-processing, first select the expected position; - data purification, Then the information is too 遽 'in the data t some records of the value of the space, some of the big or false or wrong, in the process of analysis must be excluded from these noise data 'to avoid the analysis of the results are not accurate; Through the collection of different data bases, the related rules are in line with the needs of the target problem; a code 'in the case of relevant information and complete equipment, according to the data to be mined and set the relevant numerical parameters; Exploring the relationship between the conditions of judging, judging, and qualifying, understanding the conditions of job-seeking conditions, and expecting to find useful and important potential relationships; and a report evaluation 'analysis based on association rules The characteristics of the conditions that come out, the overall requirements of the environment and the company's preferences, provide the conditions for application _ to the job seeker as a reference. The manpower database described above mines _ the connection of different types of data pastes. . And transfer, such as the 5 human resources database described in the human resources database to lie down, the original, the other process steps are mainly through different network segments, which are rich and connected communication, such as the Internet 30 201003434 :_w(Extranet) 'She is still in the form of intercommunication, transmission, and hunting to collect the target data in real time. Dig the human stomach library to mine the lion's fortune, which is based on the data, combined with the _ rule exploration technology above, the group ===___ including, __ prospect, multi-level marriage rule only wood survey and number The association rule is explored. 5. The human resource database mining association rule and system thereof as claimed in claim 4, wherein the technology of the association law exploration mainly deals with the following two sub-problems: 1. First, find out all items with a branch greater than or equal to the minimum Zhao degree. Set, called "large itemset"; -, then, from a large project, the law of confidence that the confidence level is greater than or equal to the minimum confidence level. 'The human resources database mining association law as described in claim 4 and The system, in which the multi-level association rule explores the technical characteristics of the spirit; it uses the top-down (t〇pd〇wn) processing method, first calculates the large-scale project set at level 1 〇evel-l), and then calculates the level 2 (leve 2) large project sets, and so on, until no other large project sets are produced; in each level, the Apri〇ri algorithm can be used to generate large project sets.