CN107193456A - Commending system and method based on slidingtype interactive operation - Google Patents
Commending system and method based on slidingtype interactive operation Download PDFInfo
- Publication number
- CN107193456A CN107193456A CN201710315487.6A CN201710315487A CN107193456A CN 107193456 A CN107193456 A CN 107193456A CN 201710315487 A CN201710315487 A CN 201710315487A CN 107193456 A CN107193456 A CN 107193456A
- Authority
- CN
- China
- Prior art keywords
- surfing
- stability
- user
- negative
- boundary
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
- G06F3/048—Interaction techniques based on graphical user interfaces [GUI]
- G06F3/0484—Interaction techniques based on graphical user interfaces [GUI] for the control of specific functions or operations, e.g. selecting or manipulating an object, an image or a displayed text element, setting a parameter value or selecting a range
- G06F3/04847—Interaction techniques to control parameter settings, e.g. interaction with sliders or dials
Abstract
A kind of commending system and method based on slidingtype interactive operation, comprise the following steps：1) slip information in this navigation process of monitoring user, therefrom extracts surfing stability stabtestWith surfing sequence；2) by surfing stability stabtestRespectively with stable stability boundary BpWith negative boundary of stability BnCompare, such as stabtest>BpThen recommend to browse similar content, such as stab with thistest<BnThen shielding browses similar content with this, if Bn≤stabtest≤Bp, then next step is carried out；3) cutting dimensionality reduction is carried out to surfing sequence, utilize the grader prediction user interest that scores, according to user interest to user's content recommendation, the present invention is further extracted to the hiding feature of user's surfing, it is effective to improve the accuracy recommended, interests change of the user in navigation process can effectively be judged, so as to more accurately and timely judge the final level of interest to content of user.
Description
Technical field
The present invention relates to a kind of technology in intelligent interaction field, specifically a kind of pushing away based on slidingtype interactive operation
Recommend system and method.
Background technology
Recommendation method main function based on man-machine interactive operation on intelligent terminal is can be from all kinds of man-machine interactive operations
It is middle to extract implicit user to the fancy grade of content, brought so as to make up most of users and not scored after navigation process terminates
Sparse Problem, wherein it is relatively broad be employed be exactly access frequency and read the residence time.Access frequency is to utilize
What user was recommended the access times of content, i.e., more access times mean that user is bigger to the interest of content.
It is then the average speed that make use of user in navigation process to read the residence time, i.e., the lower average speed that browses represents user
Interest to content is big.
The content of the invention
The present invention is used as implicit feedback for prior art is more by page residence time or number of clicks, and it recommends accurate
Property more low defect, propose a kind of commending system and method based on slidingtype interactive operation.
The present invention is achieved by the following technical solutions：
The present invention relates to a kind of commending system based on slidingtype interactive operation, including：Characteristic extracting module, model training
Module, score in predicting module and recommending module, wherein：Model training module determines stable stability boundary B according to existing datap
With negative boundary of stability Bn, and train scoring grader, slip information of the characteristic extracting module from user this navigation process
Extract surfing stability and surfing sequence, score in predicting module receives the surfing stability that this browses and clear
Look at velocity series and according to the stable stability boundary B received from model training modulep, negative boundary of stability BnWith scoring grader
To obtain user interest, recommending module receives user interest and generates recommendation list.
The present invention relates to the implementation method of said system, comprise the following steps：
1) slip information in this navigation process of monitoring user, therefrom extracts surfing stability stabtestWith it is clear
Look at velocity series；
2) by surfing stability stabtestRespectively with stable stability boundary BpWith negative boundary of stability BnCompare, such as
stabtest>BpThen recommend to browse similar content, such as stab with thistest<BnThen shielding browses similar content with this, such as
Fruit Bn≤stabtest≤Bp, then next step is carried out；
3) cutting dimensionality reduction is carried out to surfing sequence, using the grader prediction user interest that scores, according to user interest
To user's content recommendation.
The different numbereds of described surfing stability all sliding speed frequencies of occurrences when being browsed for user.
Described surfing sequence is one M × M speed transfer matrix, and M is the sliding speed number in navigation process,
The i-th row j column elements are by sliding speed v in matrixiTo sliding speed vjThe number of times of transfer.
Described stable stability boundary BpFor by existing positive sample set IrAfter being arranged according to surfing stability descending
All positive degree of aliasing are more than the maximum of surfing stability in the sample for just obscuring threshold value.
Described negative boundary of stability BnFor by existing negative sample set InrArranged according to surfing stability ascending order
All negative degree of aliasing are more than the minimum value of surfing stability in the negative sample for obscuring threshold value afterwards.
The cutting of described surfing sequence refers to the row and column of surfing sequence respectively according to quick-negative speed
Degree, at a slow speed-negative velocity, at a slow speed-positive speed and quick-positive speed are divided into 16 submatrixs.
Described surfing sequence carries out dimensionality reduction using PCA.
Described scoring grader is trained using support vector machine method and obtained.
Technique effect
Compared with prior art, the present invention is further extracted to the hiding feature of user's surfing, and effective improve pushes away
The accuracy recommended, can effectively judge interests change of the user in navigation process, so as to more accurately and timely judge user
The final level of interest to content.
Brief description of the drawings
Fig. 1 is schematic flow sheet of the present invention；
Fig. 2 is the surfing stability schematic diagram of positive negative sample；
Fig. 3 is the cumulative probability density map of the surfing stability of positive negative sample；
Fig. 4 is the differentiation result figure of surfing sequence；
Fig. 5 is that velocity space cutting is schematic diagram；
Fig. 6 is surfing sequence cutting schematic diagram；
Fig. 7 is recommendation accuracy rate schematic diagram；
Fig. 8 schemes for the ROC and AUC of the present invention；
Fig. 9 is the Performance Evaluation figure for combining ICF；
Figure 10 is the Performance Evaluation figure for combining SVD++.
Embodiment
The present embodiment is related to a kind of commending system based on slidingtype interactive operation, including：Characteristic extracting module, model instruction
Practice module, score in predicting module and recommending module, wherein：Model training module determines stable stability border according to existing data
BpWith negative boundary of stability Bn, and train scoring grader, slip information of the characteristic extracting module from user this navigation process
Extract surfing stability and surfing sequence, score in predicting module receives the surfing stability that this browses and clear
Look at velocity series and according to the stable stability boundary B received from model training modulep, negative boundary of stability BnWith scoring grader
To obtain user interest, recommending module receives user interest and generates recommendation list.
As shown in figure 1, the implementation of said system, comprises the following steps：
1) slip information in this navigation process of monitoring user, therefrom extracts surfing stability stabtestWith it is clear
Look at velocity series.
Described surfing stability is the difference that user is applicable all sliding speed frequencies of occurrences when touch-screen is browsed
Numbered.As shown in Fig. 2 showing that the surfing of a positive sample is distributed the surfing point with a negative sample in figure
Cloth.Positive sample refers to the interesting content browsed of user, and negative sample refers to the not interested content browsed of user.In a sample
Navigation process in there is NpIndividual different sliding speed, the frequency that each sliding speed occurs is
Then therefrom the number of statistics different frequency value is used as surfing stability.As shown in figure 3, surfing stability is less than 4
Sample be all negative sample, and the sample that surfing stability is more than 14 is all positive sample.Meanwhile, observe the accumulative general of them
Rate density is it can be found that the sample number that surfing stability is less than 4 occupies the 50% of all negative sample numbers, and surfing is steady
Fixed sample number of the degree more than 14 occupies the 30% of all positive sample numbers.
Described surfing sequence is one M × M speed transfer matrix, and M is the sliding speed number in navigation process,
The i-th row j column elements are by sliding speed v in matrixiTo sliding speed vjThe number of times of transfer.As shown in figure 4, from surfing sequence
Result after row dimensionality reduction finds out that most of positive negative sample can be distinguished by decision boundary.
2) by surfing stability stabtestRespectively with stable stability boundary BpWith negative boundary of stability BnCompare, such as
stabtest>BpThen recommend to browse similar content, such as stab with thistest<BnThen shielding browses similar content with this, such as
Fruit Bn≤stabtest≤Bp, then next step is carried out.
Described stable stability boundary BpFor by existing positive sample set IrAfter being arranged according to surfing stability descending
All positive degree of aliasing are more than the maximum of surfing stability in the sample for just obscuring threshold value.First, by positive sample set Ir
Arranged, and should be met after sequence according to surfing stability descendingThen find
One maximum surfing stability so that positive degree of aliasing, which is more than, just obscures threshold value, i.e. pcr(m)>Tpcr, this is maximum to browse
Velocity-stabilization degree is used as stable stability boundary Bp.Positive degree of aliasingWherein：Set PS(m)For IrIn it is clear
The velocity-stabilization degree stab that lookes at is more than m sample, NS(m)For negative sample set InrMiddle surfing stability stab is less than m sample
This, m is any surfing stability.
Described negative boundary of stability BnFor by existing negative sample set InrArranged according to surfing stability ascending order
All negative degree of aliasing are more than the minimum value of surfing stability in the negative sample for obscuring threshold value afterwards.First, by negative sample set
InrArranged, that is, should be met after arranging according to surfing stability ascending orderConnect
One minimum surfing stability of searching so that negative degree of aliasing ncr(m)Obscure threshold value T more than negativencr, i.e. ncr(m)>
Tncr, minimum surfing stability is negative boundary of stability Bn.Negative degree of aliasing
3) cutting dimensionality reduction is carried out to surfing sequence, using the grader prediction user interest that scores, according to user interest
Recommend to user.
As shown in Fig. 5~6, described surfing sequence carries out cutting and referred to：Each sliding speed is converted into [- 1,
Sliding speed, i.e., be converted into the relative velocity of [- 1,1] by the 1] relative velocity in scope, and interval [- 1,1] is the velocity space.Phase
In [- 1, -0.5] be quick-negative velocity to speed, it is interval (- 0.5,0] relative velocity be at a slow speed-negative velocity, it is interval (0,
0.5] relative velocity is at a slow speed-positive speed, interval (0.5,1] relative velocity be quick-positive speed.By surfing sequence
Row and column according to aforementioned four interval division be 16 pieces, and be classified as 4 classes i.e. quick-quick (FF), quick-(FS), slow at a slow speed
Speed-quick (SF) and at a slow speed-at a slow speed (SS).
Scoring grader, i.e. the surfing sequence to data with existing, which are obtained, using the training of existing data carries out cutting drop
Dimension, recycles the support vector machine method training in machine learning to obtain scoring grader.Adopted after the user interest speculated
With collaborative filtering to user's content recommendation.
As shown in Figure 7, it can be seen that under different intelligent terminal screen sizes, this method is attained by more than 80%
Accuracy rate.As screen size increases to 9.7 inches from 4.8 inches, accuracy rate falls below 83.87% from 93.08%.This be because
To increase with screen, the slip number of times needed for user has browsed full content tails off, and then result in the decline of accuracy rate.But
It is that can see, even minimum accuracy rate also has 83.87%.The influence of this account for screen size is simultaneously little.
As shown in Figure 8, it is shown that ROC images, ROC is receiver operating characteristic.It can be seen that in different screen sizes
Under, performance of the invention is all much larger than the performance of stochastic prediction algorithm.Their corresponding AUC are respectively 0.9304,0.9099,
0.8856 and 0.8383.These AUC are very close to ideal value 1, and AUC is area under receiver operator curve, and TPR is inspection
Survey accuracy.
As shown in Fig. 9~10, it can be seen that after being combined with collaborative filtering (ICF) method, common OR (optimistic estimate), NR
(estimating naturally) method is compared, and 16.35% is improved in terms of hit rate, and then improves 40.32% in hit sequence.With it is strange
Different value is decomposed after the combination of (SVD++) method, it can be seen that compared with common OR and NR, improved in terms of hit rate
23.69%th, 42.07%, then improve 14.70%, 33.87% in terms of sequence is hit.
Compared with prior art, the present invention is further extracted to the hiding feature of user's surfing, and effective improve pushes away
The accuracy recommended, can effectively judge interests change of the user in navigation process, so as to more accurately and timely judge user
The final level of interest to content.
Above-mentioned specific implementation can by those skilled in the art on the premise of without departing substantially from the principle of the invention and objective with difference
Mode local directed complete set is carried out to it, protection scope of the present invention is defined by claims and not by above-mentioned specific implementation institute
Limit, each implementation in the range of it is by the constraint of the present invention.
Claims (7)
1. a kind of commending system based on slidingtype interactive operation, it is characterised in that including：Characteristic extracting module, model training
Module, score in predicting module and recommending module, wherein：Model training module determines stable stability boundary B according to existing datap
With negative boundary of stability Bn, and train scoring grader, slip information of the characteristic extracting module from user this navigation process
Extract surfing stability and surfing sequence, score in predicting module receives the surfing stability that this browses and clear
Look at velocity series and according to the stable stability boundary B received from model training modulep, negative boundary of stability BnWith scoring grader
To obtain user interest, recommending module receives user interest and generates recommendation list.
2. a kind of recommendation method based on system described in claim 1, it is characterised in that comprise the following steps：
1) slip information in this navigation process of monitoring user, therefrom extracts surfing stability stabtestAnd surfing
Sequence；
2) by surfing stability stabtestRespectively with stable stability boundary BpWith negative boundary of stability BnCompare, such as stabtest
>BpThen recommend to browse similar content, such as stab with thistest<BnThen shielding browses similar content with this, if Bn≤
stabtest≤Bp, then next step is carried out；
3) to surfing sequence carry out cutting dimensionality reduction, using score grader prediction user interest, according to user interest to
Family content recommendation；
The different numbereds of described surfing stability all sliding speed frequencies of occurrences when being browsed for user；
Described surfing sequence is one M × M speed transfer matrix, and M is the sliding speed number in navigation process, matrix
In the i-th row j column elements be by sliding speed viTo sliding speed vjThe number of times of transfer.
3. method according to claim 1 or 2, it is characterized in that, described stable stability boundary BpFor by existing positive sample
Set IrAll positive degree of aliasing are more than surfing in the sample for just obscuring threshold value after being arranged according to surfing stability descending
The maximum of stability.
4. the recommendation method according to claim 1 or 2 based on slidingtype interactive operation, it is characterized in that, described is negative steady
Qualitative boundary BnFor by existing negative sample set InrAll negative degree of aliasing are more than after being arranged according to surfing stability ascending order
The minimum value of surfing stability in the negative sample for obscuring threshold value.
5. the recommendation method according to claim 2 based on slidingtype interactive operation, it is characterized in that, described surfing
The cutting of sequence refers to the row and column of surfing sequence respectively according to quick-negative velocity, at a slow speed-negative velocity, at a slow speed-positive speed
Degree and quick-positive speed are divided into 16 submatrixs.
6. method according to claim 1 or 2, it is characterized in that, described surfing sequence uses PCA
Carry out dimensionality reduction.
7. method according to claim 1 or 2, it is characterized in that, described scoring grader uses support vector machine method
Training is obtained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710315487.6A CN107193456B (en) | 2017-05-08 | 2017-05-08 | Commending system and method based on slidingtype interactive operation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710315487.6A CN107193456B (en) | 2017-05-08 | 2017-05-08 | Commending system and method based on slidingtype interactive operation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107193456A true CN107193456A (en) | 2017-09-22 |
| CN107193456B CN107193456B (en) | 2018-05-08 |
Family
ID=59872409
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710315487.6A Active CN107193456B (en) | 2017-05-08 | 2017-05-08 | Commending system and method based on slidingtype interactive operation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107193456B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107609786A (en) * | 2017-09-26 | 2018-01-19 | 宿州学院 | A kind of method that user behavior preference variation model is built under online community network |
| CN107909498A (en) * | 2017-10-26 | 2018-04-13 | 厦门理工学院 | Based on the recommendation method for maximizing receiver operating characteristic curve area under |
| CN109800353A (en) * | 2019-01-04 | 2019-05-24 | 上海上湖信息技术有限公司 | A kind of method and system of the real-time recommendation based on user browsing behavior |
| CN114475256A (en) * | 2022-03-11 | 2022-05-13 | 中国第一汽车股份有限公司 | Method and device for predicting motor over-temperature in electric automobile |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2000324198A (en) * | 1999-05-14 | 2000-11-24 | Nec Corp | Variable data rate transmitting device and variable data rate transmitting method |
| US20080126175A1 (en) * | 2006-11-29 | 2008-05-29 | Yahoo, Inc. | Interactive user interface for collecting and processing nomenclature and placement metrics for website design |
| CN102254034A (en) * | 2011-08-08 | 2011-11-23 | 浙江鸿程计算机系统有限公司 | Online analytical processing (OLAP) query log mining and recommending method based on efficient mining of frequent closed sequences (BIDE) |
| CN102523511A (en) * | 2011-11-09 | 2012-06-27 | 中国传媒大学 | Network program aggregation and recommendation system and network program aggregation and recommendation method |
| CN103930871A (en) * | 2011-05-09 | 2014-07-16 | 谷歌公司 | Recommending applications for mobile devices based on installation histories |
| CN104166668A (en) * | 2014-06-09 | 2014-11-26 | 南京邮电大学 | News recommendation system and method based on FOLFM model |
| CN105210069A (en) * | 2013-03-15 | 2015-12-30 | 谷歌公司 | Subscription-based mobile reading platform |
-
2017
- 2017-05-08 CN CN201710315487.6A patent/CN107193456B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2000324198A (en) * | 1999-05-14 | 2000-11-24 | Nec Corp | Variable data rate transmitting device and variable data rate transmitting method |
| US20080126175A1 (en) * | 2006-11-29 | 2008-05-29 | Yahoo, Inc. | Interactive user interface for collecting and processing nomenclature and placement metrics for website design |
| CN103930871A (en) * | 2011-05-09 | 2014-07-16 | 谷歌公司 | Recommending applications for mobile devices based on installation histories |
| CN102254034A (en) * | 2011-08-08 | 2011-11-23 | 浙江鸿程计算机系统有限公司 | Online analytical processing (OLAP) query log mining and recommending method based on efficient mining of frequent closed sequences (BIDE) |
| CN102523511A (en) * | 2011-11-09 | 2012-06-27 | 中国传媒大学 | Network program aggregation and recommendation system and network program aggregation and recommendation method |
| CN105210069A (en) * | 2013-03-15 | 2015-12-30 | 谷歌公司 | Subscription-based mobile reading platform |
| CN104166668A (en) * | 2014-06-09 | 2014-11-26 | 南京邮电大学 | News recommendation system and method based on FOLFM model |
Non-Patent Citations (2)
| Title |
|---|
| 余肖生: "基于网络用户信息行为的个性化推荐模型", 《重庆理工大学学报（自然科学）》 * |
| 王义: "基于用户行为的个性化推荐系统的设计与应用", 《计算机系统应用》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107609786A (en) * | 2017-09-26 | 2018-01-19 | 宿州学院 | A kind of method that user behavior preference variation model is built under online community network |
| CN107609786B (en) * | 2017-09-26 | 2021-02-09 | 宿州学院 | Method for constructing user behavior preference change model under online social network |
| CN107909498A (en) * | 2017-10-26 | 2018-04-13 | 厦门理工学院 | Based on the recommendation method for maximizing receiver operating characteristic curve area under |
| CN107909498B (en) * | 2017-10-26 | 2020-07-28 | 厦门理工学院 | Recommendation method based on area below maximized receiver operation characteristic curve |
| CN109800353A (en) * | 2019-01-04 | 2019-05-24 | 上海上湖信息技术有限公司 | A kind of method and system of the real-time recommendation based on user browsing behavior |
| CN114475256A (en) * | 2022-03-11 | 2022-05-13 | 中国第一汽车股份有限公司 | Method and device for predicting motor over-temperature in electric automobile |
| CN114475256B (en) * | 2022-03-11 | 2024-03-19 | 中国第一汽车股份有限公司 | Method and device for predicting motor over-temperature in electric automobile |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107193456B (en) | 2018-05-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107193456B (en) | Commending system and method based on slidingtype interactive operation | |
| EP2080113B1 (en) | Media material analysis of continuing article portions | |
| CN103440640B (en) | A kind of video scene cluster and browsing method | |
| CN101251855B (en) | Equipment, system and method for cleaning internet web page | |
| US20150066934A1 (en) | Automatic classification of segmented portions of web pages | |
| US20140143259A1 (en) | Identification and Ranking of News Stories of Interest | |
| CN105095187A (en) | Search intention identification method and device | |
| AU2016273851A1 (en) | Accurate tag relevance prediction for image search | |
| CN102135967A (en) | Webpage keywords extracting method, device and system | |
| CN105512285B (en) | Adaptive network reptile method based on machine learning | |
| CN103984959A (en) | Data-driven and task-driven image classification method | |
| CN111382605B (en) | Video content auditing method, device, storage medium and computer equipment | |
| CN105069072A (en) | Emotional analysis based mixed user scoring information recommendation method and apparatus | |
| CN103020618A (en) | Detection method and detection system for video image text | |
| CN105320778A (en) | Commodity labeling method suitable for electronic commerce Chinese website | |
| US20110010363A1 (en) | Information processing apparatus, information processing method, and program | |
| CN103425755A (en) | Method based on microblogs for analyzing audience rating situations of television programs | |
| CN103577478A (en) | Web page pushing method and system | |
| CN103020066A (en) | Method and device for recognizing search demand | |
| CN104850617A (en) | Short text processing method and apparatus | |
| CN103810162A (en) | Method and system for recommending network information | |
| CN101515272A (en) | Method and device for extracting webpage content | |
| CN104915422A (en) | Webpage collecting method and device based on browser | |
| CN106919699A (en) | A kind of recommendation method for personalized information towards large-scale consumer | |
| CN103218420B (en) | A kind of web page title extracting method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |