CN105138653B - It is a kind of that method and its recommendation apparatus are recommended based on typical degree and the topic of difficulty - Google Patents
It is a kind of that method and its recommendation apparatus are recommended based on typical degree and the topic of difficulty Download PDFInfo
- Publication number
- CN105138653B CN105138653B CN201510540419.0A CN201510540419A CN105138653B CN 105138653 B CN105138653 B CN 105138653B CN 201510540419 A CN201510540419 A CN 201510540419A CN 105138653 B CN105138653 B CN 105138653B
- Authority
- CN
- China
- Prior art keywords
- topic
- user
- target user
- difficulty
- type
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 42
- HUTDUHSNJYTCAR-UHFFFAOYSA-N ancymidol Chemical compound C1=CC(OC)=CC=C1C(O)(C=1C=NC=NC=1)C1CC1 HUTDUHSNJYTCAR-UHFFFAOYSA-N 0.000 claims abstract description 19
- 238000002474 experimental method Methods 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 4
- 230000000694 effects Effects 0.000 abstract description 11
- 238000001914 filtration Methods 0.000 description 12
- 238000004364 calculation method Methods 0.000 description 7
- 238000010586 diagram Methods 0.000 description 6
- 238000012360 testing method Methods 0.000 description 5
- 238000000205 computational method Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 241001269238 Data Species 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 238000007418 data mining Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000000630 rising effect Effects 0.000 description 1
- 239000000523 sample Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/20—Education
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Tourism & Hospitality (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Educational Administration (AREA)
- Educational Technology (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Economics (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a kind of based on typical degree and the topic of difficulty recommendation method, and method includes:It calculates each target user and does topic situation on each type topic, including:Percent of pass of quantity, difficulty and the target user of topic on this type topic;According to target user's feature vector, the similitude between arbitrary target user characteristics vector is calculated, and to each target user, select the highest user of several similarities as arest neighbors;Topic situation is done according to nearest neighbor, and prediction target user is not to doing the scoring of topic;To target user, recommendation results of several highest target topics of scoring as target user are selected.Device includes:Computing module, first choice module, grading module and the second selecting module, the present invention is realized has the characteristics that " difficulty " to improve conventional recommendation method using topic, obtains preferable effect.The present invention can be applied in study website in the future, and user is helped to select learning Content, formulate individualized learning scheme.
Description
Technical field
The present invention relates to data mining, machine learning and information retrieval fields, are related to the recommendation field of collaborative filtering, especially
It is related to a kind of based on typical degree and the topic of difficulty recommendation method and its recommendation apparatus.
Background technology
Currently, having had the commending system and proposed algorithm of comparative maturity, the proposed algorithm of current main-stream is broadly divided into association
Same filtered recommendation (collaborative filtering, CF), content-based recommendation (content based, CB) and mixing
Recommendation method (hybrid methods).Mixing proposed algorithm integrates the above two, reaches better effect.
In content-based recommendation system, article can be described as a series of vector of attribute values, and describe article
The attribute value of feature is referred to as " content ".Content-based recommendation system is exactly to find to use according to the scoring behavior of the history of user
Family preference, and recommend the article close with its preference.This method is mainly used for the article for recommending to describe with text, than
Such as:Documents and materials, news etc..
Collaborative filtering predicts user to unknown material by the similitude between similitude between user or article
The scoring of product.The arest neighbors of similitude being referred to as based on user is recommended between Main Basiss user;Phase between Main Basiss article
Recommend like the arest neighbors of degree being referred to as based on article.Collaborative filtering recommending method set up this assumes that user interest love
It remains unchanged for a long period of time well.
Whether content-based recommendation system or traditional Collaborative Filtering Recommendation System, apply it to knowledge and push away
There is its deficiency when recommending field.Content-based recommendation system is wanted accurately describe the feature of recommended article, and will
It is mapped with user preference, when facing similar to the more fuzzy data of the such feature of topic, is just difficult to accurate description, with
Cause to recommend inaccurate.Traditional collaborative filtering method does not need accurate description article, but needs a large amount of score data, when by its
When being applied to recommendation knowledge, recommending topic this aspect, it is difficult to there is enough score datas.
Invention content
Recommend method and its recommendation apparatus, the present invention can based on typical degree and the topic of difficulty the present invention provides a kind of
Effectively overcome conventional recommendation technology when being applied in knowledge recommendation, article characteristics are difficult to describe, score information is few and not fully
Consider the technical matters of this important feature of item difficulty, it is described below:
It is a kind of to recommend method, the topic that method is recommended to include the following steps based on typical degree and the topic of difficulty:
It calculates each target user and does topic situation on each type topic, including:Quantity, difficulty and the mesh of topic
Mark percent of pass of the user on this type topic;
According to target user's feature vector, the similitude between arbitrary target user characteristics vector is calculated, and to each mesh
User is marked, selects the highest user of several similarities as arest neighbors;
Topic situation is done according to nearest neighbor, and prediction target user is not to doing the scoring of topic;
To target user, recommendation results of several highest target topics of scoring as target user are selected.
Wherein, the topic recommendation method further includes:
Debug or invalid data, then sort by submission time from small to large;
The user and topic that statistical data is related to form user's set and a topic set.
Wherein, target user's feature vector is specially:
< type1:typicality1,type2:typicality2,…,typei:typicalityi…,typen:
typicalityn>
Wherein, typeiRepresent topic types;typicalityiIt is target user in typeiTypical degree on type topic;
I numbers for topic types;N is topic types sum.
A kind of topic recommendation apparatus based on typical degree and difficulty, the topic recommendation apparatus include:
Computing module does topic situation for calculating each target user on each type topic, including:The number of topic
Amount, the percent of pass of difficulty and target user on this type topic;
First choice module, for according to target user's feature vector, calculating between arbitrary target user characteristics vector
Similitude, and to each target user, select the highest user of several similarities as arest neighbors;
Grading module, for doing topic situation according to nearest neighbor, prediction target user is not to doing the scoring of topic;
Second selecting module, for target user, selecting several highest target topics of scoring as target user's
Recommendation results.
Wherein, the topic recommendation apparatus further includes:
Preprocessing module is used for debug or invalid data, then sorts from small to large by submission time;Statistical number
According to the user and topic being related to, user's set and a topic set are formed.
The advantageous effect of technical solution provided by the invention is:
1, the present invention provides new approaches for application of the commending system in user's learning process, in traditional user characteristics
It is introduced in expression and does topic difficulty to represent its ability level, improve conventional recommendation method to a certain extent in topic recommendation
Effect.
2, it is compared with the coordination filter method based on user, the present invention can preferably describe user characteristics, obtain more acurrate
Topic recommendation results, can help user select learning Content, improve learning efficiency.
Description of the drawings
Fig. 1 is a kind of flow chart for recommending method based on typical degree and the topic of difficulty;
Fig. 2 is the influence schematic diagram that degree-of-difficulty factor introduces ratio;
Fig. 3 is the comparison schematic diagram of Distance conformability degree;
Fig. 4 is the comparison schematic diagram of cosine similarity;
Fig. 5 is a kind of structural schematic diagram of the topic recommendation apparatus based on typical degree and difficulty;
Fig. 6 is a kind of another structural schematic diagram of the topic recommendation apparatus based on typical degree and difficulty.
In attached drawing, parts list represented by the reference numerals are as follows:
1:Computing module; 2:First choice module;
3:Grading module; 4:Second selecting module;
5:Preprocessing module.
Specific implementation mode
To make the object, technical solutions and advantages of the present invention clearer, embodiment of the present invention is made below further
It is described in detail on ground.For the convenience of description, the user that be recommended topic in the embodiment of the present invention is referred to as target user, title will be pushed away
The entitled target topic together recommended.
Embodiment 1
Method is recommended based on typical degree and the topic of difficulty an embodiment of the present invention provides a kind of, referring to Fig. 1, which pushes away
The method of recommending includes the following steps:
101:Preprocessed data;
The primary data of processing of the embodiment of the present invention is that user does the submission record inscribed.Debug or invalid (example first
Such as:Lack critical data item) data, then sort from small to large by submission time.After completing above-mentioned steps, statistical data relates to
And user and topic, form user set and a topic set.
102:Classify to topic set, calculation question purpose difficulty, to generate problem description;
In embodiments of the present invention, using the type label of topic (such as " Dynamic Programming ", " computational geometry ", " number of combinations
Learn " ...) classify to topic set.In addition, also to calculate the difficulty of per pass topic.In embodiments of the present invention, topic
Difficulty be described as being submitted the function of two factors of number and percent of pass, and the two factors can be by step 101
Obtained data statistics obtains.After known to the difficulty and type of topic, by per pass problem description at two tuples:<Type,
Difficulty>.
103:It calculates each target user and does topic situation, that is, the number of done this type topic on each type topic
Amount, the percent of pass of difficulty and the user on this type topic;
Topic situation is done according to above-mentioned, calculates typical degree of the target user on this type topic;Target user is each
Typical degree on type topic forms the feature description (i.e. feature vector) of the target user:
< type1:typicality1,type2:typicality2,…,typei:typicalityi…,typen:
typicalityn>
Wherein, typeiRepresent topic types, typicalityiIt is the target user in typeiTypical case on type topic
Degree, i number for topic types;N is topic types sum.
104:According to target user's feature vector, the similitude between arbitrary target user characteristics vector is calculated, and to every
A target user selects the highest user of several similarities as its " arest neighbors ";
The user characteristics formed first with step 103 are vectorial, between calculating any two target user's feature vector
Similarity, in this, as the similarity between two target users;Then to each target user, the highest use of several similarities is selected
Family is as its " arest neighbors "." arest neighbors " herein, i.e. arest neighbors described in tradition collaborative filtering method, will be used as prediction
The Main Basiss that target user scores to target topic.
105:Topic situation is done according to nearest neighbor, and prediction target user is not to doing the scoring of topic.
In embodiments of the present invention, for target topic and target user, the arest neighbors use for having done this target topic is calculated
Amount mesh is normalized to [0,1] and is scored afterwards the prediction of this target topic as user.
106:To target user, select several highest target topics of scoring as the recommendation results to the target user.
The step 106 is according to step 105 and step 103.Target user does that the frequency inscribed is higher, recommends it number of topic
Amount is more, and user does the frequency inscribed and obtained by step 103.It determines after recommending number, the pre- test and appraisal generated according to step 105
Point, select the topic of respective numbers best result as recommendation results.
The present invention is using the last submission time of user in experimental data as boundary, target user in a period of time after acquisition
Topic situation is done, by comparing recommendation results with actually doing topic situation, evaluates the accuracy of recommendation.
In conclusion the embodiment of the present invention has fully considered this important spy of item difficulty by step 101- steps 106
Sign, improves conventional recommendation method, obtains preferable effect.
Embodiment 2
Technical solution in embodiment 1 is described in detail with reference to specific calculation formula, example, it is as detailed below:
201:Preprocessed data;
That is, some illegal data are excluded, such as:Some have lacked the data of critical data item;In addition, statistics is formed just
The user's set and topic set of beginning, the data area handled with the clear embodiment of the present invention.
202:Classify to topic set, calculation question purpose difficulty;
In general, the topic done by more people is simpler, the higher topic of percent of pass is simpler, and both this compares, the former is heavier
It wants, because some topic percent of pass are very high, but only one or two people submit, and this topic is often and remarkable.In addition to this, different
It is variant between the topic of classification, such as:The topic of computational geometry class, since the size of code solved a problem is big, detailed problem is more, causes
The percent of pass of such topic entirety is not high.The degree-of-difficulty factor of in summary these actual conditions, topic is calculated by formula (1).
Wherein, j is topic piAffiliated topic class number;mxSubjIt is to be submitted the most topic of number in j class topics
Purpose submits number;mxAcjIt is the percent of pass of the highest topic of percent of pass in j class topics.subCntiAnd acRateiGeneration respectively
Table title mesh piBe submitted number and percent of pass.The item difficulty coefficient calculated by above formula belongs to [0,1].
203:Calculate user characteristics vector;
Typical degree of the user characteristics vector, that is, user on every class topic.
Computational methods (Cai, Yi, Leung, Ho-fung, Li Q, et al.TyCo according to TyCo:Towards
Typicality-based Collaborative Filtering Recommendation[C]//201022nd
International Conference on Tools with Artificial IntelligenceIEEE Computer
Society,2010:97-104), typical degree of the user in a kind of topic corresponds to user group depends on two factors, and one is
User is to the average score of this type topic, the other is user does the frequent degree of this kind of topic.User comments such topic
It is point higher, then it is interested in this kind of topic;It is more frequent that user does such topic, then interested in such topic.According to
Upper two factors calculate separately out two values, and typical degree of the user in such topic corresponds to user group will be the flat of the two values
Mean value.User uiTypical degree v in jth class topic corresponds to user groupI, jCalculation formula is as shown in Equation 2.
Wherein,It is user uiTo the average score of done jth class topic;Si,jIt is user uiDo the quantity of j class topics;
Si,kFor user uiDo the quantity of k class topics;N is topic types sum.
In fact, in the application scenarios, user to per pass topic only by with not by two kinds as a result, passing through scoring
It is 1, is otherwise 0;Therefore, average score R of the user to every class topici,jAll it is 1, the preference or energy of user can not be represented
Power is horizontal.
The application scenarios of TyCo mainly predict that film scores, and (DF_TyCo) method master that the embodiment of the present invention proposes
It is used for topic recommendation.Difficulty is an important feature specific to topic, the difficulty of the done topic of user, to a certain extent
Represent the ability level of user.User typical case's degree calculation formula is as shown in Equation 3 in the embodiment of the present invention.
Wherein, n is topic classification sum;Di,jFor user uiDo the item difficulty summation of jth class topic;Si,jFor user ui
Do the quantity of jth class topic;β is undetermined coefficient, is determined by experiment optimal value;Si,kFor user uiDo the number of kth class topic
Amount;Di,kFor user uiDo the item difficulty summation of kth class topic.
Formula 2 eliminates average score factor compared with formula 3, increases item difficulty factor, is DF_TyCo opposite
Recommend the improvement of application aspect in topic in TyCo.
204:Calculate user's similarity;
The step calculates the similarity between arbitrary two users using user characteristics vector.The embodiment of the present invention is adopted respectively
With cosine similarity formula and Distance conformability degree formula, and pass through its effect of Experimental comparison.Two in calculating formula of similarity
A input vector is the feature vector of two users respectively, i.e.,:WithWherein, vi,kFor user uiTypical degree on kth class topic;vj,kFor user ujIn kth
Typical degree on class topic.After input vector determines, the specific calculating process of similarity is known to those skilled in the art, this
Inventive embodiments do not repeat this.
205:Select arest neighbors;
After having found out the similarity between user, chooses several users most like with it for each user and form the use
" arest neighbors " at family.The embodiment of the present invention is that each user chooses the similar users of fixed quantity as its arest neighbors.Arest neighbors
User is excessive or the very few effect that can all influence to recommend, in the embodiment of the present invention by multiple contrast experiment select one it is optimal
Value, the embodiment of the present invention are not limited the number of arest neighbors.
206:Predict that user scores to topic;
Target user calculates the scoring of target topic according to its arest neighbors.What the nearest neighbor of target user was done
Topic, target user may also can do, and more similar user, and doing topic situation more has reference value.The embodiment of the present invention
Think, arbitrary user it is submitted and by topic scoring be 1, the scoring of remaining topic is 0, then with nearest neighbor and
Similarity between target user is weights, does weighted average, obtained result is prediction of the target user to target topic
Scoring.
207:Topic is recommended in selection.
Herein there are two types of method, one is one threshold value of selection, target user is scored above the topic of threshold value as pushing away
It recommends as a result, another kind is the topic for recommending fixed quantity for each target user.In the embodiment of the present invention, it is adopted as each target
It is illustrated for the topic of user's recommendation fixed quantity.
By experimental data according to the time, it is divided into test set and training set, is generated using the data in training set and recommends knot
Data comparison in fruit, with test set weighs the accuracy of recommendation.
In conclusion 201- steps 207 have the characteristics that " difficulty " embodiment of the present invention using topic through the above steps
Conventional recommendation method is improved, obtains preferable effect.The embodiment of the present invention can be applied in study website in the future, help
User selects learning Content, formulates personalized Learning Scheme.
Embodiment 3
Feasibility is carried out with reference to specific example, calculation formula, attached drawing to the technical solution in embodiment 1,2 to test
Card, it is described below:
In experiment, the value of the number K of nearest neighbor is allowed to be respectively 5,10 ..., 70, observation experiment is as a result, to probe into K values
Influence to experimental result;Under certain K values, changes user and do weight factor of the topic difficulty when describing user characteristics, see
It examines experimental result and calculates its accuracy rate.
The present invention is using F values (F-measure) and user's accuracy rate (PU) come evaluation experimental result.For convenience,
Claim " to user uxRecommend one of topic py" it is primary recommend.
If the recommendation sum generated is recomNum, the number accurately recommended is accuate, and all users do in test set
Topic sum is realSum, then accuracy pred and recall rate recall computational methods respectively as formula (4), formula (5) are shown.
There is pred and recall that can calculate F values, shown in the computational methods such as formula (6) of F values.
Wherein, pred and recall reflects the quality of result in terms of two respectively, and F values are that the comprehensive of the two embodies,
In experiment, F values are bigger, as a result better.
User's accuracy rate refers to the percentage that the target user that can be accurately recommended accounts for all target users.To each target
User, which fixes, recommends 10 topics, if for some target user, at least one of topic is to recommend accurately, then claiming should
Target user is accurately recommended.Shown in user's accuracy rate calculation formula such as formula (7):
Wherein, predUser is the number of users accurately recommended, and allUser is the total number of users being recommended.
As shown in Fig. 2, user characteristics are described as be in the typical degree on every a kind of topic.Introduce degree-of-difficulty factor, i.e., with
The difficulty of the done topic in family is a part for user characteristics.Figure it is seen that experimental result can be improved by introducing difficulty, say
It is bright in recommendation, introduce item difficulty, be reasonable.The experimental results showed that as weight factor from 0 gradually increases to 1.0, use
The trend fallen after rising is presented in family accuracy rate, and especially when weight factor is 0.85 or so, i.e. user does topic difficulty and occupies family spy
Sign 85%, remaining factor when accounting for 15% effect it is best, to prove under this application scenarios, introduce item difficulty, Neng Gougai
The effect of kind conventional recommendation.
TyCo is introduced into commending system based on the collaborative filtering method of typical degree, and conventional recommendation can be improved
The experimental result of method.In the embodiment of the present invention, on this basis, item difficulty is introduced, is further improved commending system
Experimental result when being applied to during topic is recommended.
In fact, being compared with UBCF methods, TyCo introduces the concept of typical degree, improves traditional collaborative filtering side
Method, and the embodiment of the present invention introduces item difficulty on the basis of TyCo, improves result again.Fig. 3 and Fig. 4 difference
It illustrates under different calculating formula of similarity, the embodiment of the present invention and collaborative filtering method (TyCo) based on typical degree,
The comparison of collaborative filtering method (UBCF) based on user, from figure, it can be seen that when K takes each value, this method ()
Better result can be obtained compared with other methods.It is both other to prove that this method is better than in accuracy.
Embodiment 4
A kind of topic recommendation apparatus based on typical degree and difficulty, referring to Fig. 5, which includes:
Computing module 1 does topic situation for calculating each target user on each type topic, including:Topic
The percent of pass of quantity, difficulty and target user on this type topic;
First choice module 2, for according to target user's feature vector, calculating between arbitrary target user characteristics vector
Similitude, and to each target user, select the highest user of several similarities as arest neighbors;
Grading module 3, for doing topic situation according to nearest neighbor, prediction target user is not to doing the scoring of topic;
Second selecting module 4, for target user, selecting several highest target topics of scoring as target user's
Recommendation results.
Wherein, referring to Fig. 6, which further includes:
Preprocessing module 5 is used for debug or invalid data, then sorts from small to large by submission time;Statistics
The user and topic that data are related to form user's set and a topic set.
The embodiment of the present invention, which is realized by above-mentioned module using topic, has the characteristics that " difficulty " to improve conventional recommendation
Method obtains preferable effect.The embodiment of the present invention can be applied in study website in the future, and user is helped to select in study
Hold, formulates personalized Learning Scheme.
To the model of each device in addition to doing specified otherwise, the model of other devices is not limited the embodiment of the present invention,
As long as the device of above-mentioned function can be completed.
It will be appreciated by those skilled in the art that attached drawing is the schematic diagram of a preferred embodiment, the embodiments of the present invention
Serial number is for illustration only, can not represent the quality of embodiment.
The foregoing is merely presently preferred embodiments of the present invention, is not intended to limit the invention, it is all the present invention spirit and
Within principle, any modification, equivalent replacement, improvement and so on should all be included in the protection scope of the present invention.
Claims (4)
1. a kind of recommending method based on typical degree and the topic of difficulty, which is characterized in that it includes following that the topic, which recommends method,
Step:
It calculates each target user and does topic situation on each type topic, including:Quantity, difficulty and the target of topic are used
Percent of pass of the family on this type topic;
According to target user's feature vector, the similitude between arbitrary target user characteristics vector is calculated, and use each target
Family selects the highest user of several similarities as arest neighbors;
Topic situation is done according to nearest neighbor, and prediction target user is not to doing the scoring of topic;
To target user, recommendation results of several highest target topics of scoring as target user are selected;
Wherein, target user's feature vector is specially:Typical degree of the target user on every class topic;
< type1:typicality1,type2:typicality2,…,typei:typicalityi…,typen:
typicalityn>
Wherein, typeiRepresent topic types;typicalityiIt is target user in typeiTypical degree on type topic;I is
Topic types are numbered;N is topic types sum;
The specific calculating of above-mentioned typical case's degree is as follows:
Wherein, n is topic classification sum;Di,jFor user uiDo the item difficulty summation of jth class topic;Si,jFor user uiDo jth
The quantity of class topic;β is undetermined coefficient, is determined by experiment optimal value;Si,kFor user uiDo the quantity of kth class topic;Di,k
For user uiDo the item difficulty summation of kth class topic.
2. according to claim 1 a kind of based on typical degree and the topic of difficulty recommendation method, which is characterized in that the topic
Mesh recommend method further include:
Debug or invalid data, then sort by submission time from small to large;
The user and topic that statistical data is related to form user's set and a topic set.
3. a kind of for recommending method based on typical degree and the topic of difficulty described in any claim in claim 1-2
Recommendation apparatus, which is characterized in that the recommendation apparatus includes:
Computing module does topic situation for calculating each target user on each type topic, including:The quantity of topic,
The percent of pass of difficulty and target user on this type topic;
First choice module, it is similar between arbitrary target user characteristics vector for according to target user's feature vector, calculating
Property, and to each target user, select the highest user of several similarities as arest neighbors;
Grading module, for doing topic situation according to nearest neighbor, prediction target user is not to doing the scoring of topic;
Second selecting module, for target user, selecting recommendation of several highest target topics of scoring as target user
As a result.
4. recommendation apparatus according to claim 3, which is characterized in that the recommendation apparatus further includes:
Preprocessing module is used for debug or invalid data, then sorts from small to large by submission time;Statistical data relates to
And user and topic, form user set and a topic set.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510540419.0A CN105138653B (en) | 2015-08-28 | 2015-08-28 | It is a kind of that method and its recommendation apparatus are recommended based on typical degree and the topic of difficulty |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510540419.0A CN105138653B (en) | 2015-08-28 | 2015-08-28 | It is a kind of that method and its recommendation apparatus are recommended based on typical degree and the topic of difficulty |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105138653A CN105138653A (en) | 2015-12-09 |
CN105138653B true CN105138653B (en) | 2018-08-21 |
Family
ID=54724001
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510540419.0A Active CN105138653B (en) | 2015-08-28 | 2015-08-28 | It is a kind of that method and its recommendation apparatus are recommended based on typical degree and the topic of difficulty |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105138653B (en) |
Families Citing this family (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105975564B (en) * | 2016-04-29 | 2019-07-02 | 天津大学 | Knowledge recommendation method based on opposite Entropy conformability degree |
CN105930319A (en) * | 2016-05-09 | 2016-09-07 | 北京新唐思创教育科技有限公司 | Methods and devices for establishing question knowledge point obtaining model and obtaining question knowledge point |
CN106168975B (en) * | 2016-07-12 | 2019-09-13 | 精硕科技(北京)股份有限公司 | The acquisition methods and device of target user's concentration |
CN106651696B (en) * | 2016-11-16 | 2020-10-27 | 福建天泉教育科技有限公司 | Approximate question pushing method and system |
CN109086939A (en) * | 2018-08-14 | 2018-12-25 | 合肥知多少教育科技有限公司 | A kind of input method and device doing topic result |
CN109635291B (en) * | 2018-12-04 | 2023-04-25 | 重庆理工大学 | Recommendation method for fusing scoring information and article content based on collaborative training |
CN111353015B (en) * | 2018-12-24 | 2024-03-15 | 阿里巴巴集团控股有限公司 | Crowd-sourced question recommendation method, device, equipment and storage medium |
CN110737771B (en) * | 2019-09-12 | 2022-09-27 | 北京十分科技有限公司 | Topic distribution method and device based on big data |
CN110598112A (en) * | 2019-09-17 | 2019-12-20 | 广东小天才科技有限公司 | Topic recommendation method and device, terminal equipment and storage medium |
CN110704510A (en) * | 2019-10-12 | 2020-01-17 | 中森云链(成都)科技有限责任公司 | User portrait combined question recommendation method and system |
CN110992223A (en) * | 2019-11-20 | 2020-04-10 | 大连理工大学 | Topic recommendation method of online evaluation system |
CN112100341B (en) * | 2020-04-13 | 2023-07-07 | 上海擅择教育科技有限公司 | Intelligent question classification and recommendation method for rapid expressive force test |
CN111767464A (en) * | 2020-07-09 | 2020-10-13 | 海口科博瑞信息科技有限公司 | Course platform content recommendation method, device, equipment and storage medium |
CN112232610B (en) * | 2020-12-18 | 2021-03-19 | 北京几原科技有限责任公司 | Personalized question recommendation method and system using machine learning model |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP2199963A1 (en) * | 2008-12-19 | 2010-06-23 | Xerox Corporation | System and method for recommending educational resources |
CN104239969A (en) * | 2014-09-04 | 2014-12-24 | 上海合煦信息科技有限公司 | Evaluation and problem recommendation system for individualized education |
CN104834729A (en) * | 2015-05-14 | 2015-08-12 | 百度在线网络技术(北京)有限公司 | Title recommendation method and title recommendation device |
-
2015
- 2015-08-28 CN CN201510540419.0A patent/CN105138653B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP2199963A1 (en) * | 2008-12-19 | 2010-06-23 | Xerox Corporation | System and method for recommending educational resources |
CN104239969A (en) * | 2014-09-04 | 2014-12-24 | 上海合煦信息科技有限公司 | Evaluation and problem recommendation system for individualized education |
CN104834729A (en) * | 2015-05-14 | 2015-08-12 | 百度在线网络技术(北京)有限公司 | Title recommendation method and title recommendation device |
Non-Patent Citations (1)
Title |
---|
TyCo: Towards Typicality-based Collaborative Filtering Recommendation;Yi Cai等;《22nd International Conference on Tools with Artificial Intelligence》;20101231;参见第99-100页 * |
Also Published As
Publication number | Publication date |
---|---|
CN105138653A (en) | 2015-12-09 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105138653B (en) | It is a kind of that method and its recommendation apparatus are recommended based on typical degree and the topic of difficulty | |
Mladenić | Feature subset selection in text-learning | |
Weismayer et al. | Identifying emerging research fields: a longitudinal latent semantic keyword analysis | |
Zhang et al. | Detecting spammer groups from product reviews: a partially supervised learning model | |
CN103365997B (en) | A kind of opining mining method based on integrated study | |
CN106156372B (en) | A kind of classification method and device of internet site | |
CN103106262B (en) | The method and apparatus that document classification, supporting vector machine model generate | |
CN105787068B (en) | The academic recommended method and system analyzed based on citation network and user's proficiency | |
CN103020066B (en) | A kind of method and apparatus identifying search need | |
CN104750844A (en) | Method and device for generating text characteristic vectors based on TF-IGM, method and device for classifying texts | |
CN104915447A (en) | Method and device for tracing hot topics and confirming keywords | |
CN112035658B (en) | Enterprise public opinion monitoring method based on deep learning | |
CN104392006B (en) | A kind of event query processing method and processing device | |
US20180341686A1 (en) | System and method for data search based on top-to-bottom similarity analysis | |
CN105975596A (en) | Query expansion method and system of search engine | |
CN103559622A (en) | Characteristic-based collaborative filtering recommendation method | |
CN106610970A (en) | Collaborative filtering-based content recommendation system and method | |
CN104850537B (en) | The method and device screened to content of text | |
CN109816015B (en) | Recommendation method and system based on material data | |
CN113722478B (en) | Multi-dimensional feature fusion similar event calculation method and system and electronic equipment | |
CN106445994A (en) | Mixed algorithm-based web page classification method and apparatus | |
CN108228612B (en) | Method and device for extracting network event keywords and emotional tendency | |
CN103020249A (en) | Classifier construction method and device as well as Chinese text sentiment classification method and system | |
Song et al. | An effective query recommendation approach using semantic strategies for intelligent information retrieval | |
CN107341199A (en) | A kind of recommendation method based on documentation & info general model |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |