CN105786983A - Employee individualized-learning recommendation method based on learning map and collaborative filtering - Google Patents

Abstract

The invention discloses an employee individualized-learning recommendation method based on a learning map and collaborative filtering.The method includes the steps that resource features and employee attributes are respectively extracted according to the learning resource content of an online learning platform and the practical conditions of learners (enterprise employees), a mathematical model is built, a recommendation list is calculated according to similarity to generate the recommendation result, the feedback conditions of the learners are collected to be used for improving similarity calculation, and the recommendation process is optimized.The method has certain universality in recommending the content of semi-structured data, unstructured data and multimedia learning resources, the learning map of the employees and collaborative filtering are combined, the recommendation effect is corrected and optimized, the sparse scoring matrix and learning resource recommendation, namely, cold starting, of new employees can be effectively achieved, pushing of the learning content of the online learning platform is more user-friendly, the enterprise employees are effectively assisted in rapid growing, employee training and learning cost is saved, and employee learning efficiency is improved.

Description

A kind of staff ' s individuality chemistry based on Learning Map with collaborative filtering practises recommendation method

Technical field

The present invention relates to networking technology area, specifically a kind of staff ' s individuality chemistry based on Learning Map with collaborative filtering practises recommendation method.

Background technology

On-line study (e-learning) has become one of enterprise's effective means carrying out staffs training.At present, many enterprises have built on-line study and have supported system.Along with on-line study resource become increasingly abundant with learner learning demand more complicated and various, the time that employee needs cost more consults in platform with energy and retrieves the education resource meeting oneself needs, even can not find the resource meeting own interests with positions demand, the learning demand that employee is potential on the other hand is also difficult to be found.And the mode that traditional on-line study system pushes education resource exists bigger defect, therefore there is researcher to use for reference commending system successful experience in e-commerce field, the methods such as collaborative filtering are applied in on-line study system.But the problems such as the rating matrix that collaborative filtering method itself has is sparse, cold start-up still exist in study commending system.

Summary of the invention

It is an object of the invention to provide a kind of staff ' s individuality chemistry based on Learning Map with collaborative filtering and practise recommendation method, Learning Map refers to a series of learning activity designed with ability development path and occupational planning for main shaft, is that employee learns directly embodying of development path in enterprise.Collaborative filtering is based on employee's scoring to study project by calculating the method that similarity is recommended, and both combinations can play good complementary action, optimizes individualized learning recommendation process, with the problem solving to propose in above-mentioned background technology.

For achieving the above object, the present invention provides following technical scheme:

A kind of staff ' s individuality chemistry based on Learning Map with collaborative filtering practises recommendation method, specifically comprises the following steps that

(1) Learning Map modeling

Similarity for ease of counter's work Learning Map, the position level concrete according to enterprise and employee's ability label and existing online course learning content, obtain item attribute data from data base and employee's attribute data is stored in the database table of correspondence, adopting binary coding to set up employee's Learning Map matrix model, its matrix form is

(2) Similarity Measure

1. counter's work Learning Map similarity, definition employee k is α with the Learning Map similarity between employee l₁, calculating side's formula is

${\mathrm{\α}}_1=\frac{1}{2}sim(k,l);$

Wherein, 0 < k < N, 0 < l < N, 0 < α₁< 1/2, I_kAnd I_lBeing the Learning Map matrix of employee k and employee l respectively, final similarity is employee Learning Map similarity sim (k, 0.5 times l)；

2. according to the similarity α between employee rating matrix computational item k and sundry item l to education resource project₂Computing formula is as follows:

${\mathrm{\&alpha;}}_2={\mathrm{sim}}_I(k,l)=\frac{{\mathrm{\&Sigma;}}_{r_u\&Element;u_{kl}}(r_{uk}-{\stackrel{\&OverBar;}{r}}_u)(r_{ul}-{\stackrel{\&OverBar;}{r}}_u)}{\sqrt{{\mathrm{\&Sigma;}}_{r_u\&Element;u_{kl}}{(r_{uk}-{\stackrel{\&OverBar;}{r}}_u)}^2}\sqrt{{\mathrm{\&Sigma;}}_{r_u\&Element;u_{kl}}{(r_{ul}-{\stackrel{\&OverBar;}{r}}_u)}^2}}$

Wherein, 0 < k, l < N, u_klFor giving the public employee collection of two project scorings, r simultaneously_uk, r_ulRepresent the employee u scoring to project k and l respectively,It it is employee's average score to project；

Utilize Pearson correlation coefficient counter similarity β break₂,

${\mathrm{\&beta;}}_2={\mathrm{sim}}_U(a,b)=\frac{{\mathrm{\&Sigma;}}_{r_u\&Element;u_{ab}}(r_{ai}-{\stackrel{\&OverBar;}{r}}_a)(r_{bi}-{\stackrel{\&OverBar;}{r}}_b)}{\sqrt{{\mathrm{\&Sigma;}}_{r_u\&Element;u_{ab}}{(r_{ai}-{\stackrel{\&OverBar;}{r}}_a)}^2}\sqrt{{\mathrm{\&Sigma;}}_{r_u\&Element;u_{ab}}{(r_{bi}-{\stackrel{\&OverBar;}{r}}_b)}^2}}$

Wherein, u_abCollect for public scoring employee, r_ai、r_biIt is employee a and the b scoring to same project i respectively, Represent the average score of employee；

3. similarity between similarity and final employee is calculated between final education resource project, the similarity of employee's Learning Map and be positively related based on similarity between the employee of employee's scoring, final calculating formula of similarity is,

Education resource item similarity: α=α₂

Similarity β=α between employee₁+β₂；

(3) recommendation list is built

Respectively project and employee are ranked up according to similarity, select top-k respective item and employee as neighbours, for target employee u_cThe resource item set I of visit study in each project I_k, 0 < k < N, according to similarity by I_kAll neighbours project I_lSet is included into candidate items set,

$I_k^{\&prime;}=\left(\underset{I_k}{\&cup;}{I}_l\right)-I_k$

For candidate items set I '_kInStatistics I '_kiWith I_kCumulative similarity $sim(I_{ki}^{\&prime;},I_k)={\mathrm{\&Sigma;}}_{I_k\&Element;I}sim(I_{ki}^{\&prime;},I);$

According to all candidate items I '_kiCumulative similarity sim (I '_ki, I_k), n that selects cumulative similarity maximum does not access project and constitutes orderly top-n recommendation list 1；

Top-k is selected as employee neighbours, non-scoring item to be carried out score in predicting according to the similarity 3. calculated in step (2), then according to estimation score value order does not access project by maximum n constitutes top-n recommendation list 2；

Finally, arranging parameter n and constitute recommendation list 1 and recommendation list 2, the most recommendable project in two lists of therefrom choosing recommends target employee as consequently recommended list.

As the further scheme of the present invention: in described step (3), the score in predicting of project k is calculated by employee a by equation below,

$P(a,k)={\stackrel{\&OverBar;}{r}}_a+\frac{{\&Sigma;}_{u\&Element;u_{ab}}sim(a,u)\&times;(r_{ak}-{\stackrel{\&OverBar;}{r}}_u)}{{\&Sigma;}_{u\&Element;u_{ab}}\mathrm{si}m(a,u)}$

Wherein,It is the average score of employee a, r_akIt is a scoring to project k, u_abRepresent public scoring collection.

Compared with prior art, the invention has the beneficial effects as follows:

1, compared to traditional method, the application of collaborative filtering makes the content recommendation of individualized learning be no longer limited to textual resources, to semi-structured, unstructured data, multimedia learning resource as: animation, video, audio frequency etc. also can accurately be recommended, and have certain universality on content recommendation；

2, employee's satisfaction to resource has been reacted in the scoring of resource by employee, represent the quality of this resource in most cases, at the Collaborative Recommendation set up on score data basis, there is higher accuracy, and the method makes to share each other between similar employee resource use experience, expand its study thinking, it is recommended that more efficient；

3, employee's Learning Map is combined with collaborative filtering, revises on the one hand, optimizes recommendation results, can effectively solve the sparse education resource recommendation with new employee of rating matrix and cold start-up problem on the other hand,

4, the learning content making on-line study platform pushes more personalized, effectively helps enterprise staff to shoot up, has saved staffs training, learning cost, improve employee's learning efficiency；

5, to solve the rating matrix of collaborative filtering on the one hand sparse in cold start-up problem for this recommendation method, and combining employee specifically learns route on the other hand, it is recommended that result is more accurately more personalized.

Accompanying drawing explanation

Fig. 1 is the optimization structure chart of the present invention.

Fig. 2 is the flow chart of the present invention.

Detailed description of the invention

Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is clearly and completely described, it is clear that described embodiment is only a part of embodiment of the present invention, rather than whole embodiments.Based on the embodiment in the present invention, the every other embodiment that those of ordinary skill in the art obtain under not making creative work premise, broadly fall into the scope of protection of the invention.

Refer to Fig. 1, first according to the education resource content of on-line study platform and learner (enterprise staff) practical situation, extract resource characteristic and employee's attribute respectively, founding mathematical models, then recommendation results is produced according to Similarity Measure recommendation list, and collect the feedback of learner for improving Similarity Measure, optimize recommendation process.

Such as Fig. 2, a kind of staff ' s individuality chemistry based on Learning Map with collaborative filtering practises recommendation method, specifically comprises the following steps that

(1) Learning Map modeling

Four steps that Learning Map is drawn: status analyzing, energy force modeling, course content design and Establishing, similarity for ease of counter's work Learning Map, the position level concrete according to enterprise and employee's ability label and existing online course learning content, obtain item attribute data from data base and employee's attribute data is stored in the database table of correspondence, adopting binary coding to set up employee's Learning Map matrix model, its matrix form is

(2) Similarity Measure

1. counter's work Learning Map similarity, definition employee k is α with the Learning Map similarity between employee l₁, calculating side's formula is

${\mathrm{\&alpha;}}_1=\frac{1}{2}sim(k,l);$

Wherein, 0 < k < N, 0 < l < N, 0 < α₁< 1/2, I_kAnd I_lBeing the Learning Map matrix of employee k and employee l respectively, final similarity is employee Learning Map similarity sim (k, 0.5 times l)；

2. according to the similarity α between employee rating matrix computational item k and sundry item l to education resource project₂Computing formula is as follows:

${\mathrm{\&alpha;}}_2={\mathrm{sim}}_I(k,l)=\frac{{\mathrm{\&Sigma;}}_{r_u\&Element;u_{kl}}(r_{uk}-{\stackrel{\&OverBar;}{r}}_u)(r_{ul}-{\stackrel{\&OverBar;}{r}}_u)}{\sqrt{{\mathrm{\&Sigma;}}_{r_u\&Element;u_{kl}}{(r_{uk}-{\stackrel{\&OverBar;}{r}}_u)}^2}\sqrt{{\mathrm{\&Sigma;}}_{r_u\&Element;u_{kl}}{(r_{ul}-{\stackrel{\&OverBar;}{r}}_u)}^2}}$

Wherein, 0 < k, l < N, u_klFor giving the public employee collection of two project scorings, r simultaneously_uk, r_ulRepresent the employee u scoring to project k and l respectively,It it is employee's average score to project；

Utilize Pearson correlation coefficient counter similarity β break₂,

${\mathrm{\&beta;}}_2={\mathrm{sim}}_{1J}(a,b)=\frac{{\mathrm{\&Sigma;}}_{r_u{\mathrm{EU}}_{ab}}(r_{ai}-\stackrel{\&OverBar;}{r_a})(r_{bi}-\stackrel{\&OverBar;}{r_b})}{\sqrt{{\mathrm{\&Sigma;}}_{r_u\&Element;u_{ab}}{(r_{\stackrel{\&OverBar;}{\mathrm{\&omega;}}}-\stackrel{\&OverBar;}{r_a})}^2}\sqrt{{\mathrm{\&Sigma;}}_{r_u\&Element;u_{ab}}{(r_{b\stackrel{\&OverBar;}{1}}-\stackrel{\&OverBar;}{r_b})}^2}}$

Wherein, u_abCollect for public scoring employee, r_ai、r_biIt is employee a and the b scoring to same project i respectively, Represent the average score of employee；

3. similarity between similarity and final employee is calculated between final education resource project, the similarity of employee's Learning Map and be positively related based on similarity between the employee of employee's scoring, final calculating formula of similarity is,

Education resource item similarity: α=α₂

Similarity β=α between employee₁+β₂；

(3) recommendation list is built

Respectively project and employee are ranked up according to similarity, select top-k respective item and employee as neighbours, for target employee u_cThe resource item set I of visit study in each project I_k, 0 < k < N, according to similarity by I_kAll neighbours project I_lSet is included into candidate items set,

$I_k^{\&prime;}=\left(\underset{I_k}{\&cup;}{I}_l\right)-I_k$

For candidate items set I '_kInStatistics I '_kiWith I_kCumulative similarity $sim(I_{ki}^{\&prime;},I_k)={\mathrm{\&Sigma;}}_{I_k\&Element;I}sim(I_{ki}^{\&prime;},I);$

According to all candidate items I '_kiCumulative similarity sim (I '_ki, I_k), n that selects cumulative similarity maximum does not access project and constitutes orderly top-n recommendation list 1；

Top-k is selected as employee neighbours, non-scoring item to be carried out score in predicting according to the similarity 3. calculated in step (2), then according to estimation score value order does not access project by maximum n constitutes top-n recommendation list 2；

The score in predicting of project k is calculated by employee a by equation below,

$P(a,k)={\stackrel{\&OverBar;}{r}}_a+\frac{{\&Sigma;}_{u\&Element;u_{ab}}sim(a,u)\&times;(r_{ak}-{\stackrel{\&OverBar;}{r}}_u)}{{\&Sigma;}_{u\&Element;u_{ab}}\mathrm{si}m(a,u)}$

Wherein,It is the average score of employee a, r_akIt is a scoring to project k, u_abRepresent public scoring collection；

Finally, arranging parameter n and constitute recommendation list 1 and recommendation list 2, the most recommendable project in two lists of therefrom choosing recommends target employee as consequently recommended list.

It is pointed out that when rating matrix is very sparse, system still can produce recommendation results according to the similarity of Learning Map, therefore solves the matrix Sparse Problems of collaborative filtering to a certain extent；When, when the commending contents of new user, its Learning Map model being calculated according to information such as the posies of this employee and produce relatively accurate recommending thus solving the cold start-up problem of system accordingly.

It is obvious to a person skilled in the art that the invention is not restricted to the details of above-mentioned one exemplary embodiment, and when without departing substantially from the spirit of the present invention or basic feature, it is possible to realize the present invention in other specific forms.Therefore, no matter from which point, embodiment all should be regarded as exemplary, and be nonrestrictive, the scope of the invention rather than described above limits, it is intended that all changes in the implication of the equivalency dropping on claim and scope included in the present invention.Any accompanying drawing labelling in claim should be considered as the claim that restriction is involved.

Claims (2)

1. practise recommendation method based on the staff ' s individuality chemistry of Learning Map with collaborative filtering for one kind, it is characterised in that specifically comprise the following steps that

(1) Learning Map modeling

Similarity for ease of counter's work Learning Map, the position level concrete according to enterprise and employee's ability label and existing online course learning content, obtain item attribute data from data base and employee's attribute data is stored in the database table of correspondence, adopting binary coding to set up employee's Learning Map matrix model, its matrix form is

(2) Similarity Measure

1. counter's work Learning Map similarity, definition employee k is α with the Learning Map similarity between employee l₁, calculating side's formula is

${\mathrm{\&alpha;}}_1=\frac{1}{2}sim(k,l);$

Wherein, 0 < k < N, 0 < l < N, 0 < α₁< 1/2, I_kAnd I_lBeing the Learning Map matrix of employee k and employee l respectively, final similarity is employee Learning Map similarity sim (k, 0.5 times l)；

2. according to the similarity α between employee rating matrix computational item k and sundry item l to education resource project₂Computing formula is as follows:

${\mathrm{\&alpha;}}_2={\mathrm{sim}}_I(k,l)=\frac{{\mathrm{\&Sigma;}}_{r_u\&Element;u_{kl}}(r_{uk}-{\stackrel{\&OverBar;}{r}}_u)(r_{ul}-{\stackrel{\&OverBar;}{r}}_u)}{\sqrt{{\mathrm{\&Sigma;}}_{r_u\&Element;u_{kl}}{(r_{uk}-{\stackrel{\&OverBar;}{r}}_u)}^2}\sqrt{{\mathrm{\&Sigma;}}_{r_u\&Element;u_{kl}}{(r_{ul}-{\stackrel{\&OverBar;}{r}}_u)}^2}}$

Wherein, 0 < k, l < N, u_klFor giving the public employee collection of two project scorings, r simultaneously_uk, r_ulRepresent the employee u scoring to project k and l respectively,It it is employee's average score to project；

Utilize Pearson correlation coefficient counter similarity β break₂,

${\mathrm{\&beta;}}_2={\mathrm{sim}}_U(a,b)=\frac{{\mathrm{\&Sigma;}}_{r_u\&Element;u_{ab}}(r_{ai}-{\stackrel{\&OverBar;}{r}}_a)(r_{bi}-{\stackrel{\&OverBar;}{r}}_b)}{\sqrt{{\mathrm{\&Sigma;}}_{r_u\&Element;u_{ab}}{(r_{ai}-{\stackrel{\&OverBar;}{r}}_a)}^2}\sqrt{{\mathrm{\&Sigma;}}_{r_u\&Element;u_{ab}}{(r_{bi}-{\stackrel{\&OverBar;}{r}}_b)}^2}}$

Wherein, u_abCollect for public scoring employee, r_ai、r_biIt is employee a and the b scoring to same project i respectively, Represent the average score of employee；

3. similarity between similarity and final employee is calculated between final education resource project, the similarity of employee's Learning Map and be positively related based on similarity between the employee of employee's scoring, final calculating formula of similarity is,

Education resource item similarity: α=α₂

Similarity β=α between employee₁+β₂；

(3) recommendation list is built

Respectively project and employee are ranked up according to similarity, select top-k respective item and employee as neighbours, for target employee u_cThe resource item set I of visit study in each project I_k, 0 < k < N, according to similarity by I_kAll neighbours project I_lSet is included into candidate items set,

$I_k^{\&prime;}=\left(\underset{I_k}{\&cup;}{I}_l\right)-I_k$

For candidate items set I '_kInStatistics I '_kiWith I_kCumulative similarity $sim(I_{ki}^{\&prime;},I_k)={\mathrm{\&Sigma;}}_{I_k\&Element;I}sim(I_{ki}^{\&prime;},I);$

According to all candidate items I '_kiCumulative similarity sim (I '_ki, I_k), n that selects cumulative similarity maximum does not access project and constitutes orderly top-n recommendation list 1；

Top-k is selected as employee neighbours, non-scoring item to be carried out score in predicting according to the similarity 3. calculated in step (2), then according to estimation score value order does not access project by maximum n constitutes top-n recommendation list 2；

Finally, arranging parameter n and constitute recommendation list 1 and recommendation list 2, the most recommendable project in two lists of therefrom choosing recommends target employee as consequently recommended list.

2. the staff ' s individuality chemistry based on Learning Map with collaborative filtering according to claim 1 practises recommendation method, it is characterised in that in described step (3), the score in predicting of project k is calculated by employee a by equation below,

$P(a,k)={\stackrel{\&OverBar;}{r}}_a+\frac{{\mathrm{\&Sigma;}}_{u\&Element;u_{ab}}sim(a,u)\&times;(r_{ak}-{\stackrel{\&OverBar;}{r}}_u)}{{\mathrm{\&Sigma;}}_{u\&Element;u_{ab}}sim(a,u)}$

Wherein,It is the average score of employee a, r_akIt is a scoring to project k, u_abRepresent public scoring collection.