CN109408600A - A kind of books based on data mining recommend purchaser's method - Google Patents
A kind of books based on data mining recommend purchaser's method Download PDFInfo
- Publication number
- CN109408600A CN109408600A CN201811112444.9A CN201811112444A CN109408600A CN 109408600 A CN109408600 A CN 109408600A CN 201811112444 A CN201811112444 A CN 201811112444A CN 109408600 A CN109408600 A CN 109408600A
- Authority
- CN
- China
- Prior art keywords
- books
- cluster
- data
- initial
- book
- 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
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
A kind of Books in University Library based on reader conduct data mining recommends purchaser's method, this method enlivens books including excavation and single two processes of buying books are recommended in matching, it proposes and a kind of books method for digging is enlivened based on improved K-means algorithm, this method determines the noise spot in sample set by the way that adjustable threshold value is arranged, then the initial cluster center that sample is chosen further according to maximum distance method, is finally obtained based on the algorithm and enlivens books.It is demonstrated experimentally that comparing tradition K-means algorithm, the improved K-means algorithm stability in enlivening books mining process of the present invention is good, accuracy rate is high, meets the demand that libraries of the universities' books recommend purchase.
Description
Technical field
The present invention relates to data mining, K-means algorithm improvement, books to recommend purchase, is a kind of books based on data mining
Recommend purchaser's method.
Background technique
With current era society and politics, culture, rapid development of economy, each subject crossing fusion, Bibliographical Information is not
Disconnected growth.According to " Chinese Books retail market report in 2016 " display, 2016, national book retail market is dynamic to sell kind number
1725.09 ten thousand, new book kind number is about 21.03 ten thousand.New book kind from 2012 to 2016 year always 200,000 to 210,000 kinds it
Between.In face of the library resource of broad categories, how libraries of the universities, which choose, not only conformed to quality requirements, but also met Reader's Demand
Books will be faced with huge challenge.Traditional Literature Acquisition is the major way that Current Libraries collect that books recommend purchase information,
But which heavy workload, low efficiency are difficult to hold the changeable book need of reader in time.By consulting nearly 5 years domestic foreign ministers
Documents and materials are closed, books personalized recommendation research towards reader of the discovery based on computer-related technologies is more, towards books
It is relatively fewer that the books in shop recommend purchase research.Traditional literature interview can not adapt to Current College Library and recommend purchase demand, existing
It is less that the books supported based on computer technology recommend purchase technique study, and due to a lack of the analysis to behavioral datas such as Readers ' Borrowing Books,
It is difficult to accurately hold Reader's Demand, be not widely used by libraries of the universities.According to above-mentioned, devise a kind of based on reader conduct
The books of data mining recommend purchase model, and the model is based on reader library behavioral data, first with improved K-means
Algorithm excavation enlivens books, then segments to enlivening books, keyword therein is extracted, finally, according to these keywords
The book list in book list or recommendation system for purchase provided with bookman matches, so that intelligence quickly provides and recommends list of buying books, significantly
Improve the initiative for recommending purchase, science, accuracy and timeliness.
Summary of the invention
Purchase demand is recommended in order to overcome the shortcomings of that traditional literature interview can not adapt to Current College Library, based on calculating
It is less that the books of machine technical support recommend purchase technique study, and due to a lack of the analysis to behavioral datas such as Readers ' Borrowing Books, it is difficult to accurate
Reader's Demand is held, is not widely used by libraries of the universities, the present invention provides a kind of books based on data mining and recommends purchaser
Method, stability is good, accuracy rate is high, meets the demand that libraries of the universities' books recommend purchase.
In order to solve the above-mentioned technical problem the present invention adopts the following technical scheme:
A kind of books based on data mining recommend purchaser's method, the described method comprises the following steps:
The first step, excavation enliven books: both having been met quality by clustering scheduling algorithm based on reader conduct data and wanted
The books asked and welcome by reader;
The initial data of research includes Reader Data table readerinfo, Collection Data table colinfo, expert data table
Expscore, tables of data lendinfo and retrieval tables of data searchinfo are borrowed, by book borrowing and reading ratio, to shop duration, inspection
Rope number enlivens five publishing house, readership accounting features as evaluation and enlivens the foundations of books;
It is inputted the cluster feature set after data processing as data set, and data is normalized, will taken
Value is mapped on section [0,1];A set local noise spot is created, k are chosen in the sample set of removal noise point set
Initial cluster center calculates each sample point at a distance from initial cluster center using Euclidean distance, is divided to distance most
Cluster where small initial cluster center;Judge whether it is less than the threshold value of setting according to the size of cluster, it is if being less than that this is first
Beginning cluster centre is put into noise point set, chooses cluster centre again;Otherwise the mean value M1 for calculating each cluster, generates according to mean value
New mass center calculates each sample point further according to Euclidean distance and generates new cluster at a distance from M1, calculates the mean value M2 of new cluster, if |
M2-M1 |>1, continue to adjust cluster, if | M2-M1 | obtained cluster is output to file by<1, is obtained and is enlivened books, and algorithm terminates;
List of buying books is recommended in second step, matching: will be enlivened books with existing and is recommended single matching of buying books, obtains and finally recommends list of buying books;
Further, in the first step, the process that initial cluster center is chosen is as follows:
1.1) assume that cluster number is K, cluster centre indicates then there is A={ a with A1,a2,a3,…,ak, of sample point
Number is N, and noise spot number is O, attribute M, then sample set X={ X1,X2,X3,…,XN-O, calculate any two sample point
xi,xjBetween Euclidean distance Dij, find out maximum DIJIfThen sample point xi,xjWith regard to as initial clustering
The first two class heart, that is, have a1=xi,a2=xj;It goes to step and 2) calculates remaining K-2 cluster centre;
1.2) the first two initial cluster center of cluster had been calculated by last step, it is assumed that have determined that k at this time
Cluster centre (2≤k≤K-1), then+1 cluster centre of kth is ai+1, be exactly in sample set remaining N-O-k sample point with
The preceding k cluster centre having determined apart from minimum value Di;Maximum D is looked for againIIfThen sample point
xiK-th of the initial cluster center exactly clustered repeats step 1.2), until K initial cluster center is all found in this way;
1.3) determination of noise spot: noise spot threshold parameter is set as t, cluster is generated according to initial cluster center, is judged whether
Number in each cluster is greater than threshold value N/Kt, otherwise will be initial in this cluster if it is greater than the step then continued below algorithm
Cluster centre is labeled as noise spot, and noise spot is removed when algorithm chooses initial cluster center again, until of each cluster
Number meets the threshold value of setting.
Further, in the first step, analytical calculation is carried out to initial data, process is as follows:
1) ratio is borrowed
It is proposed is borrowed ratio and is calculated with following formula:
B in formulaRRatio, B are borrowed in expressionNNumber, C are borrowed in expressionNIndicate duplicate number;
2) to shop duration
According to book number book_id, action type operation, operating time operation_time and duplicate number copy
The average to shop duration of every book is calculated, is related to borrowing tables of data lendinfo and Collection Data table colinfo;
3) number is retrieved
Number is retrieved to be counted according to reader's number reader_id, retrieval content content, retrieval time search_time
It obtains;
4) publishing house is enlivened
Building enlivens publishing house's data mining model:
P=PA×wA+PB×wB+PC×wC
In formula, P indicates the comprehensive score of some publishing house, PA、PB、PCRespectively indicate Readers ' Borrowing Books, expert recommends, history
Scoring of the procurement information to publishing house, wA、wB、wCRespectively indicate Readers ' Borrowing Books, expert recommends, historical purchase is enlivening publishing house
Set weight in scoring excavation;Whether the publishing house that every book is judged according to the size of P value is to enliven publishing house;
5) readership accounting
Readership accounting indicates that the identity for borrowing the reader of certain this book is different, and the existing value of the style of calligraphy is also different;
R=RT×wT+RG×wG+RU×wU
In formula, the value of readership accounting, R are representedT、RG、RURespectively indicate the teacher's number for borrowing certain this book, postgraduate
Number, undergraduate's number, wT、wG、wURespectively represent teacher, postgraduate, weight shared by three kinds of identity of undergraduate.
Further, in the second step, matching recommend buy books singly be exactly enliven books and it is existing recommend buy books singly it is matched
Process segments tool first with Stanford CoreNLP, carries out word segmentation processing to books title is enlivened, title is divided into name
Word, adjective, conjunction, character and number;Then the unrelated word in removal participle, obtains title keyword;Finally, will extract
Keyword and recommending buy books it is single carry out string matching, export the book list containing keyword in title, these books list are to recommend purchase
Book list.
Technical concept of the invention are as follows: the present invention enlivens books including excavation and single two processes of buying books are recommended in matching.It proposes
It is a kind of that books method for digging is enlivened based on improved K-means algorithm, this method by be arranged adjustable threshold value come it is true
Then the noise spot of this concentration of random sample chooses the initial cluster center of sample further according to maximum distance method, be finally based on the algorithm
It obtains and enlivens books.It is demonstrated experimentally that comparing tradition K-means algorithm, improved K-means algorithm is enlivening books mining process
Middle stability is good, accuracy rate is high, meets the demand that libraries of the universities' books recommend purchase.
Beneficial effects of the present invention are mainly manifested in: 1, intelligence, which quickly provides, recommends list of buying books, and substantially increases and recommends purchase
Initiative, science, accuracy and timeliness.2, improved K-means algorithm improves accuracy rate and stability.
Detailed description of the invention
Fig. 1 is the detailed process that books recommend purchase model, and main includes that excavation enlivens books, single two mistakes of buying books are recommended in matching
Journey.
Fig. 2 is to excavate the process for enlivening books, mainly includes three data acquisition, data processing and data mining steps.
Fig. 3 illustrates table data store relationship and structure, including Reader Data table (readerinfo), Collection Data table
(colinfo), expert data table (expscore), borrow tables of data (lendinfo), retrieval tables of data (searchinfo),
In, keepers reader's reading Borrowing System such as tables of data, retrieval tables of data is borrowed, other tables constitute the complete of Readers ' Borrowing Books behavior
Whole expression.
Fig. 4 illustrates improved K-means algorithm flow.
Fig. 5 illustrates the situation of change of threshold parameter difference accuracy rate.
Specific embodiment
The invention will be further described below in conjunction with the accompanying drawings.
Referring to Fig.1~Fig. 5, a kind of books based on data mining recommend purchaser's method, include the following steps:
The first step, excavation enliven books: both having been met quality by clustering scheduling algorithm based on reader conduct data and wanted
The books asked and welcome by reader;
Firstly, the acquisition of initial data
The initial data of research is from University Library Business system and relevant automatic system, including Reader Data table
Readerinfo, Collection Data table colinfo, expert data table expscore, tables of data lendinfo, retrieval tables of data are borrowed
Searchinfo, wherein borrow keepers reader's reading Borrowing System such as tables of data, retrieval tables of data, other tables constitute reading
The expressed intact of person's Borrowing System.
Then, data processing
Reader conduct determines the liveness of books.By book borrowing and reading ratio, to shop duration, retrieval number, active publication
Five society, readership accounting features enliven main foundation and important poly- of subsequent K-means algorithm of books as evaluation
Category feature, using the maximum cluster of characteristic value after cluster as enlivening books.Obtain above-mentioned five features, need to initial data into
Row analytical calculation.
1) ratio is borrowed
It borrows number and is most intuitively demonstrated by demand of the reader to the book.But library preservation duplicate number is to borrowing number
It has a certain impact.Influence in order to avoid duplicate number difference to experiment, proposition are borrowed ratio and are calculated with following formula:
B in formulaRRatio, B are borrowed in expressionNNumber, C are borrowed in expressionNIndicate duplicate number.
2) to shop duration
Smaller expression book of value to shop duration is frequently borrowed.Mainly according to book number (book_id), action type
(operation), operating time (operation_time) and duplicate number (copy) calculate the average to shop duration of every book.It relates to
And borrow tables of data (lendinfo) and Collection Data table (colinfo).
3) number is retrieved
It is particularly significant for discovery Reader's Demand to retrieve number, the retrieval record retrieved in information table is the true of reader conduct
Real record, can objectively and comprehensively react the potential demand of reader.Number is retrieved according in reader's number (reader_id), retrieval
Hold (content), retrieval time (search_time) statistics obtains.
4) publishing house is enlivened
The publishing house that publication amount is big, quality is high, there are the books compared with high usage and larger reader's influence power is known as active
Publishing house.Books are enlivened for what is excavated, in conjunction with the demand of the students and faculty and the holdings structure in library, comprehensively consider influence
The factors of Literature Acquisition, angularly from Readers ' Borrowing Books, expert's recommendation, historical purchase, building enlivens publishing house's data mining
Model.
P=PA×wA+PB×wB+PC×wC
In formula, P indicates the comprehensive score of some publishing house, PA、PB、PCRespectively indicate Readers ' Borrowing Books, expert recommends, history
Scoring of the procurement information to publishing house, wA、wB、wCRespectively indicate Readers ' Borrowing Books, expert recommends, historical purchase is enlivening publishing house
Set weight in scoring excavation.Whether the publishing house that every book is judged according to the size of P value is to enliven publishing house.
5) readership accounting
Readership accounting indicates that the identity for borrowing the reader of certain this book is different, and the existing value of the style of calligraphy is also different.
Teacher, postgraduate, undergraduate represent three different identity, therefore give different considerations to the reader of different identity and weigh
Weight.
R=RT×wT+RG×wG+RU×wU
In formula, R represents the value of readership accounting, RT、RG、RURespectively indicate the teacher's number for borrowing certain this book, postgraduate
Number, undergraduate's number, wT、wG、wURespectively represent teacher, postgraduate, weight shared by three kinds of identity of undergraduate
Furthermore it is clustered using improved K-means clustering algorithm.
The algorithm is inputted using the cluster feature set after data processing as data set, and place is normalized to data
Reason, value is mapped on section [0,1].A set local noise spot is created, in the sample set of removal noise point set
K initial cluster center is chosen, each sample point is calculated at a distance from initial cluster center using Euclidean distance, is divided to
Cluster where the smallest initial cluster center.Judge whether it is less than the threshold value of setting according to the size of cluster, if being less than
The initial cluster center is put into noise point set, chooses cluster centre again;Otherwise the mean value M1 for calculating each cluster, according to equal
Value generates new mass center, calculates each sample point further according to Euclidean distance and generates new cluster at a distance from M1, calculates the mean value of new cluster
M2, if | M2-M1 |>1, continue to adjust cluster, if | M2-M1 | obtained cluster is output to file by<1, and algorithm terminates.
Initial cluster center selection is the key link of this algorithm, and detailed process is as follows:
1.1) assume that cluster number is K, cluster centre indicates then there is A={ a with A1,a2,a3,…,ak, of sample point
Number is N, and noise spot number is O, attribute M, then sample set X={ X1,X2,X3,…,XN-O, calculate any two sample point
xi,xjBetween Euclidean distance Dij, find out maximum DIJIfThen sample point xi,xjWith regard to as initial clustering
The first two class heart, that is, have a1=xi,a2=xj.It goes to step and 2) calculates remaining K-2 cluster centre.
1.2) the first two initial cluster center of cluster had been calculated by last step, it is assumed that have determined that k at this time
Cluster centre (2≤k≤K-1), then+1 cluster centre of kth is ai+1, be exactly in sample set remaining N-O-k sample point with
The preceding k cluster centre having determined apart from minimum value Di.Maximum D is looked for againIIfThen sample point
xiK-th of the initial cluster center exactly clustered repeats step 2), until K initial cluster center is all found in this way.
1.3) determination of noise spot: noise spot threshold parameter is set as t, cluster is generated according to initial cluster center, is judged whether
Number in each cluster is greater than threshold value N/Kt, otherwise will be initial in this cluster if it is greater than the step then continued below algorithm
Cluster centre is labeled as noise spot, and noise spot is removed when algorithm chooses initial cluster center again, until of each cluster
Number meets the threshold value of setting, just continues the step below algorithm.
List of buying books is recommended in second step, matching: will be enlivened books with existing and is recommended single matching of buying books, obtains and finally recommends list of buying books
Matching recommend buy books singly be exactly enliven books and it is existing recommend single matched process of buying books, be a kind of active, automation
It obtains and finally recommends single process of buying books.Libraries of the universities recommend list of buying books and are mainly obtained by following methods at present: 1) recommending purchase system
The book list that reader recommends in system;2) book list that librarian is obtained in a manner of questionnaire survey etc.;3) book list that bookman provides.Though
It is also it will be apparent that needs that the book list that right above-mentioned three kinds of modes provide, which has certain one-sidedness recommended purchase reasonability, but recommend purchase,
Fining selection again is carried out to book list.Matching recommends single process of buying books and segments tool first with Stanford CoreNLP[8],
Word segmentation processing is carried out to books title is enlivened, title is divided into noun, adjective, conjunction, character, number etc.;Then it removes
The unrelated words such as number, character, conjunction in participle obtain title keyword;Finally, by the keyword of extraction and recommending list of buying books
String matching is carried out, the book list containing keyword in title is exported, these books list are to recommend list of buying books.
Claims (4)
1. a kind of books based on data mining recommend purchaser's method, it is characterised in that: the described method comprises the following steps:
The first step, excavation enliven books: based on reader conduct data, by clustering scheduling algorithm, not only conformed to quality requirements but also
The books welcome by reader;
The initial data of research includes Reader Data table readerinfo, Collection Data table colinfo, expert data table
Expscore, tables of data lendinfo and retrieval tables of data searchinfo are borrowed, by book borrowing and reading ratio, to shop duration, inspection
Rope number enlivens five publishing house, readership accounting features as evaluation and enlivens the foundations of books;
It is inputted the cluster feature set after data processing as data set, and data is normalized, value is reflected
It penetrates on section [0,1];A set local noise spot is created, k are chosen in the sample set of removal noise point set initially
Cluster centre calculates each sample point at a distance from initial cluster center using Euclidean distance, is divided to apart from the smallest
Cluster where initial cluster center;Judge whether it is less than the threshold value of setting according to the size of cluster, it is if being less than that this is initial poly-
Class center is put into noise point set, chooses cluster centre again;Otherwise the mean value M1 for calculating each cluster is generated newly according to mean value
Mass center calculates each sample point further according to Euclidean distance and generates new cluster at a distance from M1, calculates the mean value M2 of new cluster, if | M2-M1
|>1, continue to adjust cluster, if | M2-M1 | obtained cluster is output to file by<1, is obtained and is enlivened books, and algorithm terminates;
List of buying books is recommended in second step, matching: will be enlivened books with existing and is recommended single matching of buying books, obtains and finally recommends list of buying books.
2. a kind of books based on data mining as described in claim 1 recommend purchaser's method, it is characterised in that: the first step
In, the process that initial cluster center is chosen is as follows:
1.1) assume that cluster number is K, cluster centre indicates then there is A={ a with A1,a2,a3,…,ak, the number of sample point is
N, noise spot number are O, attribute M, then sample set X={ X1,X2,X3,…,XN-O, calculate any two sample point xi,xj
Between Euclidean distance Dij, find out maximum DIJIfThen sample point xi,xjBefore as initial clustering
Two class hearts, that is, have a1=xi,a2=xj;It goes to step and 2) calculates remaining K-2 cluster centre;
1.2) the first two initial cluster center of cluster had been calculated by last step, it is assumed that have determined that k cluster at this time
Center (2≤k≤K-1), then+1 cluster centre of kth is ai+1, be exactly in sample set remaining N-O-k sample point with
Determining preceding k cluster centre apart from minimum value Di;Maximum D is looked for againIIfThen sample point xiJust
It is k-th of initial cluster center of cluster, step 1.2) is repeated in this way, until K initial cluster center is all found;
1.3) determination of noise spot: noise spot threshold parameter is set as t, cluster is generated according to initial cluster center, is judged whether each
Number in cluster is greater than threshold value N/Kt, if it is greater than the step then continued below algorithm, otherwise by the initial clustering in this cluster
Centre mark is noise spot, and noise spot is removed when algorithm chooses initial cluster center again, until the number of each cluster is full
The threshold value set enough.
3. a kind of books based on data mining as claimed in claim 1 or 2 recommend purchaser's method, it is characterised in that: described first
In step, analytical calculation is carried out to initial data, process is as follows:
1) ratio is borrowed
It is proposed is borrowed ratio and is calculated with following formula:
B in formulaRRatio, B are borrowed in expressionNNumber, C are borrowed in expressionNIndicate duplicate number;
2) to shop duration
It is calculated according to book number book_id, action type operation, operating time operation_time and duplicate number copy
Every book it is average to shop duration, be related to borrowing tables of data lendinfo and Collection Data table colinfo;
3) number is retrieved
Retrieval number is obtained according to reader's number reader_id, retrieval content content, retrieval time search_time statistics
?;
4) publishing house is enlivened
Building enlivens publishing house's data mining model:
P=PA×wA+PB×wB+PC×wC
In formula, P indicates the comprehensive score of some publishing house, PA、PB、PCRespectively indicate Readers ' Borrowing Books, expert recommends, historical purchase
Scoring of the information to publishing house, wA、wB、wCRespectively indicate Readers ' Borrowing Books, expert recommends, historical purchase is enlivening publishing house's scoring
Set weight in excavation;Whether the publishing house that every book is judged according to the size of P value is to enliven publishing house;
5) readership accounting
Readership accounting indicates that the identity for borrowing the reader of certain this book is different, and the existing value of the style of calligraphy is also different;
R=RT×wT+RG×wG+RU×wU
In formula, the value of readership accounting, R are representedT、RG、RURespectively indicate the teacher's number for borrowing certain this book, postgraduate's number,
Undergraduate's number, wT、wG、wURespectively represent teacher, postgraduate, weight shared by three kinds of identity of undergraduate.
4. a kind of books based on data mining as claimed in claim 1 or 2 recommend purchaser's method, it is characterised in that: described second
In step, it is singly exactly to enliven books to recommend single matched process of buying books with existing that matching, which is recommended and bought books, first with Stanford
CoreNLP segments tool, carries out word segmentation processing to books title is enlivened, title is divided into noun, adjective, conjunction, character
And number;Then the unrelated word in removal participle, obtains title keyword;Finally, by the keyword of extraction and recommending buy books it is single into
Line character String matching, exports the book list containing keyword in title, these books list are to recommend list of buying books.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811112444.9A CN109408600B (en) | 2018-09-25 | 2018-09-25 | Book recommendation method based on data mining |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811112444.9A CN109408600B (en) | 2018-09-25 | 2018-09-25 | Book recommendation method based on data mining |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109408600A true CN109408600A (en) | 2019-03-01 |
CN109408600B CN109408600B (en) | 2022-04-08 |
Family
ID=65466288
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811112444.9A Active CN109408600B (en) | 2018-09-25 | 2018-09-25 | Book recommendation method based on data mining |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109408600B (en) |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110111227A (en) * | 2019-04-12 | 2019-08-09 | 广州图创计算机软件开发有限公司 | A kind of digital publication based on Internet of Things borrows platform |
CN110110116A (en) * | 2019-04-02 | 2019-08-09 | 浙江工业大学 | A kind of trademark image retrieval method for integrating depth convolutional network and semantic analysis |
CN110532306A (en) * | 2019-05-27 | 2019-12-03 | 浙江工业大学 | A kind of Library User's portrait model building method dividing k-means based on multi-angle of view two |
CN112035450A (en) * | 2020-07-30 | 2020-12-04 | 深圳市中盛瑞达科技有限公司 | Data warehouse real-time construction method based on button |
CN113065711A (en) * | 2021-04-13 | 2021-07-02 | 河南工程学院 | Book purchasing optimization decision system and book purchasing decision method |
CN115577152A (en) * | 2022-09-29 | 2023-01-06 | 山东开正信息产业有限公司 | Online book borrowing management system based on data analysis |
CN115797018A (en) * | 2023-01-09 | 2023-03-14 | 广东拓迪智能科技有限公司 | Book recommendation method, system and storage medium |
CN116662671A (en) * | 2023-07-24 | 2023-08-29 | 中国标准化研究院 | Digital library data pushing method based on user preference |
CN116842945A (en) * | 2023-07-07 | 2023-10-03 | 中国标准化研究院 | Digital library data mining method |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103366009A (en) * | 2013-07-24 | 2013-10-23 | 杭州东信北邮信息技术有限公司 | Book recommendation method based on self-adaption clustering |
CN103886048A (en) * | 2014-03-13 | 2014-06-25 | 浙江大学 | Cluster-based increment digital book recommendation method |
US20160092512A1 (en) * | 2014-09-26 | 2016-03-31 | Kobo Inc. | System and method for using book recognition to recommend content items for a user |
US20160171112A1 (en) * | 2014-12-16 | 2016-06-16 | Kobo Incorporated | Method and system for fastest-read category e-book recommendation |
CN108090074A (en) * | 2016-11-22 | 2018-05-29 | 上海阿法迪智能标签系统技术有限公司 | Book recommendation system and method |
-
2018
- 2018-09-25 CN CN201811112444.9A patent/CN109408600B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103366009A (en) * | 2013-07-24 | 2013-10-23 | 杭州东信北邮信息技术有限公司 | Book recommendation method based on self-adaption clustering |
CN103886048A (en) * | 2014-03-13 | 2014-06-25 | 浙江大学 | Cluster-based increment digital book recommendation method |
US20160092512A1 (en) * | 2014-09-26 | 2016-03-31 | Kobo Inc. | System and method for using book recognition to recommend content items for a user |
US20160171112A1 (en) * | 2014-12-16 | 2016-06-16 | Kobo Incorporated | Method and system for fastest-read category e-book recommendation |
CN108090074A (en) * | 2016-11-22 | 2018-05-29 | 上海阿法迪智能标签系统技术有限公司 | Book recommendation system and method |
Cited By (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110110116A (en) * | 2019-04-02 | 2019-08-09 | 浙江工业大学 | A kind of trademark image retrieval method for integrating depth convolutional network and semantic analysis |
CN110110116B (en) * | 2019-04-02 | 2021-04-06 | 浙江工业大学 | Trademark image retrieval method integrating deep convolutional network and semantic analysis |
CN110111227B (en) * | 2019-04-12 | 2020-01-21 | 广州图创计算机软件开发有限公司 | Digital book borrowing platform based on Internet of things |
CN110111227A (en) * | 2019-04-12 | 2019-08-09 | 广州图创计算机软件开发有限公司 | A kind of digital publication based on Internet of Things borrows platform |
CN110532306B (en) * | 2019-05-27 | 2021-12-17 | 浙江工业大学 | College library user portrait model construction method based on multi-view binary k-means |
CN110532306A (en) * | 2019-05-27 | 2019-12-03 | 浙江工业大学 | A kind of Library User's portrait model building method dividing k-means based on multi-angle of view two |
CN112035450A (en) * | 2020-07-30 | 2020-12-04 | 深圳市中盛瑞达科技有限公司 | Data warehouse real-time construction method based on button |
CN113065711A (en) * | 2021-04-13 | 2021-07-02 | 河南工程学院 | Book purchasing optimization decision system and book purchasing decision method |
CN113065711B (en) * | 2021-04-13 | 2023-05-12 | 河南工程学院 | Book purchasing optimization decision system and book purchasing decision method |
CN115577152A (en) * | 2022-09-29 | 2023-01-06 | 山东开正信息产业有限公司 | Online book borrowing management system based on data analysis |
CN115577152B (en) * | 2022-09-29 | 2024-01-26 | 深圳市科图自动化新技术有限公司 | Online book borrowing management system based on data analysis |
CN115797018A (en) * | 2023-01-09 | 2023-03-14 | 广东拓迪智能科技有限公司 | Book recommendation method, system and storage medium |
CN116842945A (en) * | 2023-07-07 | 2023-10-03 | 中国标准化研究院 | Digital library data mining method |
CN116662671A (en) * | 2023-07-24 | 2023-08-29 | 中国标准化研究院 | Digital library data pushing method based on user preference |
CN116662671B (en) * | 2023-07-24 | 2023-10-27 | 中国标准化研究院 | Digital library data pushing method based on user preference |
Also Published As
Publication number | Publication date |
---|---|
CN109408600B (en) | 2022-04-08 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109408600A (en) | A kind of books based on data mining recommend purchaser's method | |
Weismayer et al. | Identifying emerging research fields: a longitudinal latent semantic keyword analysis | |
Sarawagi et al. | Open-domain quantity queries on web tables: annotation, response, and consensus models | |
Tuarob et al. | Automatic tag recommendation for metadata annotation using probabilistic topic modeling | |
CN104850633B (en) | A kind of three-dimensional model searching system and method based on the segmentation of cartographical sketching component | |
US9081852B2 (en) | Recommending terms to specify ontology space | |
Deshpande et al. | Text summarization using clustering technique | |
US20130103625A1 (en) | Information processing apparatus and method, and program thereof | |
CN106339502A (en) | Modeling recommendation method based on user behavior data fragmentation cluster | |
CN103927358A (en) | Text search method and system | |
CN113962293B (en) | LightGBM classification and representation learning-based name disambiguation method and system | |
CN105975596A (en) | Query expansion method and system of search engine | |
CN104899229A (en) | Swarm intelligence based behavior clustering system | |
Schedl et al. | A music information system automatically generated via web content mining techniques | |
Kim et al. | Diversifying query suggestions based on query documents | |
CN109492022A (en) | The searching method of semantic-based improved k-means algorithm | |
CN103778206A (en) | Method for providing network service resources | |
CN110795613B (en) | Commodity searching method, device and system and electronic equipment | |
Ma et al. | A new query recommendation method supporting exploratory search based on search goal shift graphs | |
CN109145161A (en) | Chinese Place Names querying method, device and equipment | |
Wei et al. | Online education recommendation model based on user behavior data analysis | |
Yao et al. | Online deception detection refueled by real world data collection | |
CN116629258B (en) | Structured analysis method and system for judicial document based on complex information item data | |
CN102043846A (en) | Search method and device based on genetic algorithm | |
Jayashree et al. | Multimodal web page segmentation using self-organized multi-objective clustering |
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 |