CN102890713B - A kind of music recommend method based on user's current geographic position and physical environment - Google Patents
A kind of music recommend method based on user's current geographic position and physical environment Download PDFInfo
- Publication number
- CN102890713B CN102890713B CN201210349719.7A CN201210349719A CN102890713B CN 102890713 B CN102890713 B CN 102890713B CN 201210349719 A CN201210349719 A CN 201210349719A CN 102890713 B CN102890713 B CN 102890713B
- Authority
- CN
- China
- Prior art keywords
- music
- collection
- tag
- photo
- vector
- 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
Abstract
The invention discloses a kind of music recommend method based on user's current geographic position and physical environment, a kind of reasonably music recommend framework of the present invention, and devise the expression structure of a kind of user's current location and physical environment feature, the efficient Vectors matching algorithm of this structural support; Meanwhile, for space attribute and the text attribute of magnanimity photo object, present invention employs the mutation space index structure supporting efficient insertion, deletion, renewal rewards theory; Based on this index structure, the spatiality that the Semantic of the combining music lyrics and photo have and Semantic, can obtain best recommendation music for user according to the current location of oneself and physical environment.
Description
Technical field
The present invention relates to Indexing Mechanism and the querying method in spatial database field, and information retrieval field is for the retrieval recommended technology of mass multimedia object, particularly relates to a kind of music recommend method based on user's current geographic position and physical environment.
Background technology
In spatial database field, in order to access massive spatial data quickly and efficiently, experts and scholars propose a large amount of space index methods, and common indexing means comprises grid (Grid), quaternary tree (Quad-Tree), R-index, R* set index, K-D-B and set index, Hilbert curve index.On this basis, more propose the various inquiry that differs from one another and solution thereof, as NN Query, k nearest neighbor inquiry, Continuous Nearest Neighbors Inquiry, oppositely NN Query, farthest neighbor queries, skyline inquiry.These spatial indexs usually with the structure organization spatial object of hierarchical, thus support efficient space querying.Set for the R be widely adopted, the data point that spatially position is close is by cluster in minimum bounding box, and these minimum bounding boxs carry out cluster again according to spatial locality recurrence, until arrive root node.
In information retrieval field, common full-text search engine adopts inverted file (Inverted File) to carry out index to document object usually, and inverted file is using document keyword as index, and document sets up keyword-Document mapping structure as index object.When user input keyword retrieve time, the collection of document that search engine can utilize inverted file to obtain efficiently to comprise this keyword and the number of times that keyword occurs in each document, thus convenient calculating web document and user inquire about between matching degree, and return Query Result by certain putting in order.The most frequently used model of existing searching system is vector space model, and each document d is mapped as proper vector V (d)=(t in the model
1, t
2, t
3t
n), wherein t
i(i=1 ... n) being the weight of the mutually different Tag of row in document d, is TF-IDF at the Tag weighing computation method that information retrieval field is the most frequently used.Use vector space model just can calculate the matching degree between any document and user's inquiry, first user's query conversion is become to be positioned at the vector of same document space, then uses the computing functions such as such as COS distance to calculate matching degree.
In music recommend field, according to the difference of method, traditional music recommend system probably can be divided into two classes.One is based on music content (Content-Based), and the extraction of music that first this method is liked from user goes out various feature, then uses the method for machine learning to judge the fancy grade of user for other music according to these features; Another kind is collaborative filtering (Collaborative-Filtering), and it supposes that the music that similar user likes also is similar, and what therefore recommend user is the music that other have the user of similar interests and like.But most commending system does not consider the context environmental that user is residing at that time, recommend music out can not meet user's psychological needs at that time, because a lot of user is according to residing occasion, generation event or physical environment situation instead of singer etc. selected music at that time in fact.
Summary of the invention
The object of the invention is to overcome the deficiencies in the prior art, a kind of music recommend method based on user's current geographic position and physical environment is provided.
The object of the invention is to be achieved through the following technical solutions: a kind of music recommend method based on user's current geographic position and physical environment, comprises the steps:
(1) photograph collection with GPS information is crawled from photo website;
(2) music collections comprising the relevant informations such as the lyrics is crawled from music site;
(3) spatial index is set up to the GPS information photograph collection that has that step 1) obtains, in the process the photo Tag collection that word segmentation processing obtains specification is carried out to the original Tag collection often opening photo;
(4) to step 2) the relevant information music collections such as the lyrics that comprises that obtains sets up and stores index, carry out participle in the process obtain corresponding music Tag collection and corresponding music-documents vector to the lyrics of every song;
(5) analytical documentation collection obtains a Tag similarity matrix;
(6) by query steps 5) the Tag collection often opening photo that step 3) obtained of the Tag similarity matrix that obtains converts the photo document vector being positioned at music-documents vector space to;
(7) the photo document vector TF-IDF weight calculation mode that music-documents vector sum step 6) step 4) obtained obtains converts cum rights music-documents vector sum cum rights photo document vector to;
(8) client receives current location information and the physical context information of user, passes to server and calculates;
(9) server goes out by the photograph collection search index that step 3) is set up all photos that customer location comprises according to the customer position information obtained from step 8) and obtains position relevant picture collection, strengthens the cum rights Tag collection obtaining representative of consumer current physical environment to the physical context information obtained from step 8) simultaneously;
(10) the cum rights Tag set pair music of the representative of consumer physical environment using step 9) to obtain is carried out filtration and is obtained candidate music collection;
(11) the photo Tag collection that the position relevant picture collection obtained according to step 9) comprises calculates the cum rights document vector of representative of consumer position, and calculates the K song mated most with it by vector similarity computing method;
(12) the photo Tag collection that the position relevant picture collection obtained based on step 9) comprises calculates corresponding label-cloud, finally the label-cloud of the K song inquired and generation is returned to client.
The beneficial effect that the present invention has is: propose a kind of reasonably music recommend framework, and devise the expression structure of a kind of user's current location and physical environment feature, the efficient Vectors matching algorithm of this structural support.Meanwhile, for space attribute and the text attribute of magnanimity photo object, have employed the mutation space index structure supporting efficient insertion, deletion, renewal rewards theory.Based on this index structure, the spatiality that the Semantic of the combining music lyrics and photo have and Semantic, can obtain best recommendation music for user according to the current location of oneself and physical environment.
Accompanying drawing explanation
Fig. 1 is the invention process flow chart of steps;
Fig. 2 is the key map according to user's current location inquiry photograph collection.
Embodiment
Describe the present invention in detail below in conjunction with accompanying drawing, object of the present invention and effect will become more obvious.
As shown in Figure 1, the music recommend method that the present invention is based on user's current geographic position and physical environment specifically comprises following implementation step:
Step 1: crawl the photograph collection with GPS information from photo website.
Crawl the API single photo access speed too slow (crawling successively according to photo ID) that photo website in the flow process of data provides, and the demand of photo is several necessarily even more than one hundred million, the present invention adopts dichotomy to detect and optimumly once crawls number of pictures, then batch downloads photo (batch downloads all photos in certain area), greatly reduce the access of network I/O, the speed such as crawling photo from www.flickr.com can reach and crawl 2000 photos 1 second, finally data cleansing is carried out after all photos have been downloaded, ensure that often opening photo only stays a corresponding data, and weed out the photo not having gps coordinate, finally define one and have the database that more than 6,000 ten thousand have the photo of GPS information.
Step 2: crawl the music collections comprising the relevant informations such as the lyrics from music site.
For crawling of www.lyrics.com website, per song is an all corresponding html file, uses regular language to analyze html file and can obtain per song corresponding singer's name, song title and the lyrics.Use multithreading accelerates, and finally have collected the musical database that contains tens0000 songs.
Step 3: set up spatial index to the GPS information photograph collection that has that step 1 obtains, carries out to the original Tag collection often opening photo the photo Tag collection that word segmentation processing obtains specification in the process.
Because picture data amount is too huge, need first to carry out pre-service to all photos, extract gps coordinate corresponding to photo and photo ID, comment number, URL, Tag collection etc.Then be inserted into according to gps coordinate in the spatial index safeguarding all photos, such as R-TREE (other spatial indexs can be adopted), maintain in the middle of R-TREE and often open the gps coordinate of photo and the ID of photo, and all the other information of photo all leave on disk.No matter the positional information of user is expressed as rectangle or circular can travel through R-TREE node by sequence and do the data that rectangle or circular cap obtain all photos in this region.
Be described for Fig. 2.For convenience of description, each rectangle in spatial index employing R-TREE, figure is exactly the minimum area-encasing rectangle of MBR(in R-TREE level), internal node comprises each MBR in lower one deck, and recurrence comprises successively.Store the GPS geographic coordinate of all photos being positioned at leaf node MBR and corresponding photo ID in leaf node, in figure, each point represents a photo.Dashed rectangle in figure represents the position rectangle that user is current, the photo comprised in this position rectangle when inquiry constitutes the photograph collection (representing these photos with black point in figure) of representative of consumer positional information, then we gathered by the Tag this photograph collection to disk obtaining these photos and accordingly other such as comment on, uplink time etc. information.
The photo Tag that some returns is spliced by some real Tag, the Tag participle adopting AC automat (Aho-Corasick string matching algorithm) these to be spliced, obtains the Tag collection of specification corresponding to photo.That such as server returns is character string BeautifulBuildingOfUSA, and it is split as Beautiful, Building, Of, USA.
Step 4: set up storage index to the relevant information music collections such as the lyrics that comprise that step 2 obtains, carries out participle to the lyrics of every song in the process and obtains corresponding music Tag collection and corresponding music-documents vector.
Music information leaves in disk according to music ID order, open up a buffer zone in internal memory and deposit a certain proportion of continuous music information, the access of music information is according to the order of sequence, emptying buffer whenever needing the music information of accessing next part, reads in the music information of next part again from disk.
By English string segmentation technology, the lyrics of per song are carried out participle and obtain corresponding music Tag collection.
The each different Tag occurred in all wordbooks is regarded as a dimension, and per song regards a document as, and musical signification is become a vector in this hyperspace, the value in each dimension is exactly the number of times that corresponding Tag occurs in this song.
Step 5: analytical documentation collection obtains a Tag similarity matrix.
The all english articles occurred in first analytical documentation collection such as wikipedia, great Ying dictionary, with a moving window obtain Tag between symbiosis number of times, then with LINShi theoretical information value be each Tag calculate one distribution similarity vector, then the similarity matrix of a similar LSA matrix is finally produced by successive ignition, wherein every a line represents the similarity vector of a Tag and other Tag, each is exactly the similarity between two Tag, and presses descending order.Like this can fast fetching must be the most similar with each Tag Tag.
The conveniently inquiry of the most similar Tag of each Tag, some Tag that precalculated each Tag is the most similar also according to order arrangement from high to low, generate a similar rectangle, deposit in disk, be loaded in internal memory at program run duration.
Step 6: the Tag collection that step 3 is often opened photo by the Tag similarity matrix obtained by query steps 5 converts the photo document vector being positioned at music-documents vector space to.
Need to calculate the similarity between music and a photo Tag set in music matching process, but photo Tag collection and music Tag collection (obtaining after lyrics participle) are in the data in two different spaces in essence, calculate likeness in form degree just needs them to be transformed in same space to calculate.Method calculates music Tag for each Tag x of photo to concentrate the Tag y the most similar with it, and replace x with y.
If such as the Tag x occurred in photo will be transformed into a music Tag to concentrate the tag occurred, first all Tag occurred in song are put in a Hash table, then scanning the Tag the most similar to x in the data line that x is corresponding in Tag similarity matrix from front to back checks in the Hash table whether appearing at music Tag collection, if find, exit, otherwise x does not have corresponding music Tag, just need not consider yet.
After having processed, convert the Tag collection often opening photo to photo document vector that one is in music vector space according to the method similar with step 4.
Step 7: the photo document vector TF-IDF weight calculation mode that the music-documents vector sum step 6 step 4 obtained obtains converts cum rights music-documents vector sum cum rights photo document vector to.
The number (DF) of the number of times (TF) occurred in it and the document vector including it in whole document vector set calculated in document vector residing for it for any one Tag occurred in music-documents vector sum photo document vector and obtains IDF value, calculating this Tag TF-IDF value corresponding in residing document vector according to TF and IDF value.
To each Tag in each document vector according to obtaining cum rights music-documents vector sum cum rights photo document vector after above-mentioned step process.
Step 8: client receives current location information and the physical context information of user, passes to server and calculates.
User specifies the current location residing for her (can be rectangle or circle by client, length and width and radius all can be changed, the coordinate of rectangle and circle shape all represents with longitude and latitude), and select the mood (selected by word or graphically select) of oneself and represent the weighted value (1 to 100) of mood intensity, client calculates other such as weather conditions simultaneously, time etc. physical context information, these information are passed to server by client.
Step 9: server goes out by the photograph collection search index that step 3 is set up all photos that customer location comprises according to the customer position information obtained from step 8 and obtains position relevant picture collection, strengthens the cum rights Tag collection obtaining representative of consumer current physical environment to the physical context information obtained from step 8 simultaneously.
From photo spatial index, all photographic intelligence collection that customer location comprises are inquired according to the method for recursive query.
The physical environment of user comprises weather conditions, time, mood etc., wherein user mood is expressed as a Tag and corresponding weight (1 to 100), the current physical environment of weight higher expression user is stronger, and is expanded by the physical environment that WordNet-Affect is current to user: the Tag of the mood similar with user's current physical environment is added in the physical environment cum rights Tag collection of expression user.
Step 10: the cum rights Tag set pair music of the representative of consumer physical environment using step 9 to obtain is carried out filtration and obtained candidate music collection.
For each Tag that the cum rights Tag of representative of consumer physical environment concentrates, threshold value between the 0-1 going out an equal proportion according to the weight computing of its correspondence, uses the Tag similarity matrix that obtains in step 5 can in the hope of the similarity of its most similar Tag of any a piece of music neutralization and correspondence.For any a piece of music, just this song is filtered out if this cum rights Tag concentrates any one Tag Tag similarity the most similar in this song not exceed corresponding threshold value, just no longer consider in music matching process.
Step 11: the photo Tag collection that the position relevant picture collection obtained according to step 9 comprises calculates the cum rights document vector of representative of consumer position, and calculates the K song mated most with it by vector similarity computing method;
The cum rights document vector V that calculate representative of consumer position similar to the method calculating cum rights photo vector in step 7, V is the same with all music-documents vectors to be all in same vector space, Similarity Measure between any one music-documents vector M and V adopts COS distance formula, is described below: (can adopt other vector similarity computing formula):
Suppose that the vector form of V is as follows:
;
Music-documents vector M form is as follows:
;
Wherein, n is the dimension of vector.
Similarity Measure between them is as follows:
。
If user's request is the music that front K head mates most, safeguard that a capacity is the most rickle of K, the coupling mark of per song calculate complete after correspondingly upgrade the content of most rickle, even this song score is just deleted heap top music higher than the music that most rickle mid-score is minimum and inserts this song, and what finally record in most rickle is exactly the highest K song of matching degree.
Step 12: the photo Tag collection that the position relevant picture collection obtained based on step 9 comprises calculates corresponding label-cloud, finally the label-cloud of the K song inquired and generation is returned to client.
Claims (1)
1., based on a music recommend method for user's current geographic position and physical environment, it is characterized in that, comprise the steps:
(1) photograph collection with GPS information is crawled from photo website;
(2) music collections comprising the relevant informations such as the lyrics is crawled from music site;
(3) spatial index is set up to the GPS information photograph collection that has that step 1) obtains, in the process the photo Tag collection that word segmentation processing obtains specification is carried out to the original Tag collection often opening photo;
(4) to step 2) the relevant information music collections such as the lyrics that comprises that obtains sets up and stores index, carry out participle in the process obtain corresponding music Tag collection and corresponding music-documents vector to the lyrics of every song;
(5) analytical documentation collection obtains a Tag similarity matrix;
(6) by query steps 5) the Tag collection often opening photo that step 3) obtained of the Tag similarity matrix that obtains converts the photo document vector being positioned at music-documents vector space to;
(7) the photo document vector TF-IDF weight calculation mode that music-documents vector sum step 6) step 4) obtained obtains converts cum rights music-documents vector sum cum rights photo document vector to;
(8) client receives current location information and the physical context information of user, passes to server and calculates;
(9) server goes out by the photograph collection search index that step 3) is set up all photos that customer location comprises according to the customer position information obtained from step 8) and obtains position relevant picture collection, strengthens the cum rights Tag collection obtaining representative of consumer current physical environment to the physical context information obtained from step 8) simultaneously;
(10) the cum rights Tag set pair music of the representative of consumer physical environment using step 9) to obtain is carried out filtration and is obtained candidate music collection;
(11) the photo Tag collection that the position relevant picture collection obtained according to step 9) comprises calculates the cum rights document vector of representative of consumer position, and calculates the K song mated most with it by vector similarity computing method;
(12) the photo Tag collection that the position relevant picture collection obtained based on step 9) comprises calculates corresponding label-cloud, finally the label-cloud of the K song inquired and generation is returned to client.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210349719.7A CN102890713B (en) | 2012-09-20 | 2012-09-20 | A kind of music recommend method based on user's current geographic position and physical environment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210349719.7A CN102890713B (en) | 2012-09-20 | 2012-09-20 | A kind of music recommend method based on user's current geographic position and physical environment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102890713A CN102890713A (en) | 2013-01-23 |
| CN102890713B true CN102890713B (en) | 2015-08-12 |
Family
ID=47534215
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210349719.7A Active CN102890713B (en) | 2012-09-20 | 2012-09-20 | A kind of music recommend method based on user's current geographic position and physical environment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102890713B (en) |
Families Citing this family (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104142936A (en) * | 2013-05-07 | 2014-11-12 | 腾讯科技（深圳）有限公司 | Audio and video match method and audio and video match device |
| CN103365984B (en) * | 2013-07-04 | 2016-08-10 | 上海交通大学 | Obtain the terrestrial reference method and system of the most farthest single neighbours on road network |
| CN104850537B (en) * | 2014-02-17 | 2017-12-15 | 腾讯科技（深圳）有限公司 | The method and device screened to content of text |
| CN104268154A (en) * | 2014-09-02 | 2015-01-07 | 百度在线网络技术（北京）有限公司 | Recommended information providing method and device |
| CN104618446A (en) * | 2014-12-31 | 2015-05-13 | 百度在线网络技术（北京）有限公司 | Multimedia pushing implementing method and device |
| CN105975496A (en) * | 2016-04-26 | 2016-09-28 | 清华大学 | Music recommendation method and device based on context sensing |
| CN107025251B (en) * | 2016-07-29 | 2021-07-20 | 杭州网易云音乐科技有限公司 | Data pushing method and device |
| WO2018027833A1 (en) * | 2016-08-11 | 2018-02-15 | 张焰焰 | Data collection method for music playback technology, and playback system |
| CN106339492B (en) * | 2016-08-31 | 2019-05-24 | 电子科技大学 | A kind of music recommended method based on geographical location information |
| CN106339504A (en) * | 2016-09-23 | 2017-01-18 | 乐视控股（北京）有限公司 | Music pushing method and apparatus |
| CN106933955A (en) * | 2017-01-22 | 2017-07-07 | 斑马信息科技有限公司 | The vehicle-mounted content service system of adaptability and devices and methods therefor |
| CN106886575A (en) * | 2017-01-22 | 2017-06-23 | 斑马信息科技有限公司 | The vehicle-mounted content service system of adaptability and devices and methods therefor |
| CN107016062B (en) * | 2017-03-16 | 2021-02-26 | 中国联合网络通信有限公司广东省分公司 | Method and system for identifying trial listening cheating behaviors |
| CN107451302B (en) * | 2017-09-22 | 2020-08-28 | 深圳大学 | Modeling method and system based on position top-k keyword query under sliding window |
| CN107800793A (en) * | 2017-10-27 | 2018-03-13 | 江苏大学 | Car networking environment down train music active pushing system |
| CN108363769A (en) * | 2018-02-07 | 2018-08-03 | 大连大学 | The method for building up of semantic-based music retrieval data set |
| CN108334617A (en) * | 2018-02-07 | 2018-07-27 | 大连大学 | The method of semantic-based music retrieval |
| CN109255680A (en) * | 2018-09-04 | 2019-01-22 | 浙江经济职业技术学院 | A kind of accurate personalized recommendation method towards agricultural product big data |
| CN109934629A (en) * | 2019-03-12 | 2019-06-25 | 重庆金窝窝网络科技有限公司 | A kind of information-pushing method and device |
| CN111177071B (en) * | 2019-12-12 | 2023-07-07 | 广州地理研究所 | Picture downloading method and device of Flickr platform |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101464881A (en) * | 2007-12-21 | 2009-06-24 | 音乐会技术公司 | Method and system for generating media recommendations in a distributed environment based on tagging play history information with location information |
| CN101984437A (en) * | 2010-11-23 | 2011-03-09 | 亿览在线网络技术（北京）有限公司 | Music resource individual recommendation method and system thereof |
| CN102654859A (en) * | 2011-03-01 | 2012-09-05 | 北京彩云在线技术开发有限公司 | Method and system for recommending songs |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2011100182A (en) * | 2009-11-04 | 2011-05-19 | Nec Corp | Recommendation information distribution system, server, portable terminal, recommendation information distribution method, program thereof, and recording medium |
-
2012
- 2012-09-20 CN CN201210349719.7A patent/CN102890713B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101464881A (en) * | 2007-12-21 | 2009-06-24 | 音乐会技术公司 | Method and system for generating media recommendations in a distributed environment based on tagging play history information with location information |
| CN101984437A (en) * | 2010-11-23 | 2011-03-09 | 亿览在线网络技术（北京）有限公司 | Music resource individual recommendation method and system thereof |
| CN102654859A (en) * | 2011-03-01 | 2012-09-05 | 北京彩云在线技术开发有限公司 | Method and system for recommending songs |
Non-Patent Citations (2)
| Title |
|---|
| 以情景感知为基之音乐推荐;翁頌舜等;《第15届海峡两岸信息管理发展与策略学术研讨会》;20090830;496-501 * |
| 音乐情感自动分析研究;蒋盛益等;《计算机工程与设计》;20101231;第31卷(第18期);4112－4115 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102890713A (en) | 2013-01-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102890713B (en) | A kind of music recommend method based on user's current geographic position and physical environment | |
| Ji et al. | Mining city landmarks from blogs by graph modeling | |
| Liu et al. | Image retagging using collaborative tag propagation | |
| CN110263180B (en) | Intention knowledge graph generation method, intention identification method and device | |
| Qian et al. | Semantic-aware top-k spatial keyword queries | |
| JP2017157192A (en) | Method of matching between image and content item based on key word | |
| CN103455487B (en) | The extracting method and device of a kind of search term | |
| Fang et al. | Folksonomy-based visual ontology construction and its applications | |
| CN104834693A (en) | Depth-search-based visual image searching method and system thereof | |
| CN103955529A (en) | Internet information searching and aggregating presentation method | |
| JP2017157193A (en) | Method of selecting image that matches with content based on metadata of image and content | |
| Ullah et al. | Search engine optimization algorithms for page ranking: comparative study | |
| Kim et al. | TwitterTrends: a spatio-temporal trend detection and related keywords recommendation scheme | |
| Chen et al. | HIB-tree: An efficient index method for the big data analytics of large-scale human activity trajectories | |
| Pakojwar et al. | Web data extraction and alignment using tag and value similarity | |
| Feng et al. | A system for region search and exploration | |
| Richter et al. | Leveraging community metadata for multimodal image ranking | |
| Zhao et al. | Integration of link and semantic relations for information recommendation | |
| Qiu et al. | Detection and optimized disposal of near-duplicate pages | |
| Deng et al. | A novel information search and recommendation services platform based on an indexing network (short paper) | |
| Viana et al. | Semantic keyword-based retrieval of photos taken with mobile devices | |
| CN112749246B (en) | Evaluation method and device of search phrase, server and storage medium | |
| Chen et al. | A blog clustering approach based on queried keywords | |
| Lei et al. | Tag recommendation for cultural resources | |
| Duklan et al. | Classification of search engine optimization techniques: A data mining approach |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |