CN100573532C - A kind of intelligent prompt method of file label - Google Patents
A kind of intelligent prompt method of file label Download PDFInfo
- Publication number
- CN100573532C CN100573532C CNB2008101061577A CN200810106157A CN100573532C CN 100573532 C CN100573532 C CN 100573532C CN B2008101061577 A CNB2008101061577 A CN B2008101061577A CN 200810106157 A CN200810106157 A CN 200810106157A CN 100573532 C CN100573532 C CN 100573532C
- Authority
- CN
- China
- Prior art keywords
- file
- label
- user
- occurs
- span
- 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.)
- Expired - Fee Related
Links
Images
Landscapes
- User Interface Of Digital Computer (AREA)
Abstract
The invention discloses a kind of intelligent prompt method of file label, belong to computer network field.Method of the present invention by utilizing file content context and the historical data of user interactions, the user is carried out providing when label adds the prompting of intelligence.Compared with prior art, the present invention can provide the label that more accurately enriches prompting for the user.
Description
Technical field
The present invention relates in table for computer plane system or the website, the reminding method the when reminding method when file label adds, particularly photo tag add belongs to computer network field.
Background technology
In recent years, digitizing process (Cyberization) [1] is advanced by leaps and bounds, and is embodied in growing continuously and fast of aspect technology such as processor, storer, audio/video coding, shooting and display device.The technical progress in these fields makes the computer user can create, obtain and manage the information with increasing data volume.With the digital photograph is example, and along with digital camera with have mobile phone universal day by day of camera function, the quantity growth of family's digital photograph is rapid.
We notice, are that most Software tools and photo are shared the function that the website all provides for photo adds label.Existing Software tool such as ACDSee[2], Microsoft Windows Photo Gallery[3], Adobe Photoshop Album[4], Google Picasa[5], representational photo is shared website Flickr.com[6] be provided as all that photo adds label and according to the function of label search photo.Meanwhile, along with photo is shared the rise of website and the improvement at mark tagging user interface, increasing photo all has label.Before photo is shared the website rise, though a lot of Software tools also are provided as the function that photo adds label, user's enthusiasm not high [7], the investigation of Ames etc. [8] shows, share photo with other people, become the user and be willing to mean the major reason that photo adds label.Label will more and more be accepted by the user as one of principal feature of Web2.0.
In recent years, having number of research projects to concentrate on to the user adds label offers help.These research work mainly can be divided into two classes, are respectively photo classification [9-11], and the automatic interpolation of label or prompting [12-14].Wherein, the work of many photo classifications is all attempted photo according to event classification, and this is because " incident " is considered to the important clue that the user remembers family's photo.In order to reach this purpose, extract low-level image feature except the method that adopts graphical analysis, also consider time [9], place [10] and the acquisition parameters metadata such as [11] of photo.Different with photo classification, the automatic interpolation of label or prompting are when user's interpolation action takes place, and provide some label alternate items to the user, allow the user therefrom select to finish interpolation.Method among the present invention also belongs to this class.The previous work of this respect mainly concentrates on the people tag of prompting photo, will utilize the algorithm [12] [13] of people's face, the identification of clothes texture isotype usually.But algorithm for pattern recognition has limitation, and such as not being positive when people's face, when perhaps the clothes texture was not obvious, these algorithms may lose efficacy.In order to overcome this deficiency, in [14], the author proposes to utilize the context of photo content to strengthen the label accuracy of predicting, and has provided complete prompting algorithm based on this contextual photo people tag.Wherein, the context of photo content comprises the shooting time of photo, place, and the existing people tag of photo etc.
We notice that except the context of file content, user and system interaction history have also comprised the user to the abundant memory clue [15] of file.Context when people such as Karl [16] pass through the recording user operation file makes up the document classification based on task (task-based), helps the user search file.File that context during operation file comprises user's operation history, the file of opening in a period of time before, opened in a period of time afterwards or the like.Undoubtedly, these interactive history data are added label to the user also prompt facility, but seldom relates in the previous work.
In addition, what the present invention mainly discussed is the reminding method that generic-document is added label, get in touch and distinguish below the reminding method that itself and comparison film add label exists: one, photo is a kind of concrete form of generic-document, generic-document also has the context relevant with content, therefore utilize context to carry out forecast method, generic-document also is suitable for.Two, the method analyzed of the picture material of comparison film is inapplicable to generic-document.Because generic-document may not have picture material.
To sum up, the limitation of previous work is mainly reflected in following three aspects: one, the scope of Tao Luning only is confined to photo, and the label that does not have to propose generic-document is pointed out algorithm; Two, only be confined to prompting, do not have this method is extended to prompting to general label people tag; Three, the historical data of user interactions is not fully excavated, for the user provides abundanter interpolation clue.
Summary of the invention
The objective of the invention is to propose a kind of intelligent prompt method of file label at three limitation above-mentioned.This method is not only utilized the context of file content, and utilizes the historical data of user interactions, and the user is carried out label prompting (being not limited to people tag).
The present invention passes through the context of extraction document content, the interactive history of analysis user and operating system, and a series of label alternate item of generation of intelligence is as the prompting of the user being added label.
The context of file content mainly is the attribute relevant with file content.Comparison film mainly comprises time, place of photograph taking etc.
The historical data of user interactions mainly comprises the creation-time of current file, nearest modification time etc.
The basic thought of method is among the present invention:
To never adding the file of label: find out and its file, add up these file acceptances of the bid and check out existing frequency with identical or close content context, the label that frequency is high more, the possibility that occurs in current file is also big more.Find out the running time alternative document identical or close in the interactive history, add up these file acceptances of the bid and check out existing frequency with current file, the label that frequency is high more, the possibility that occurs in current file is also big more.
To the file of some labels is arranged: find out the file that has identical or close content context with it, add up the frequency that label in these files and the label in the current file occur simultaneously, the label that frequency is high more, the possibility that occurs in current file are also big more.Find out the running time alternative document identical or close in the interactive history, add up the frequency that label occurs simultaneously in the label and current file in these files with current file, the label that frequency is high more, the possibility that occurs in current file is also big more.
Technical scheme of the present invention is:
A kind of intelligent prompt method of file label, the file at not adding label the steps include:
1) from context, extracts a reference element value;
2) set a span according to the reference element value;
3) extract the file of reference element value in this span, obtain file set P
μ 'And corresponding tally set T
μ '
4) statistics tally set T
μ 'In each element at file set P
μ 'The middle times N that occurs
1
5) from user interaction history extracting data one reference data value;
6) set a span according to this reference data value;
7) extract the file of reference data value in this span, obtain file set Q and corresponding tally set T ' thereof;
8) times N that each element occurs in file set Q among the statistics tally set T '
2
9) times N that occurs according to each tag element
1And N
2Label is sorted, and the prompting user selects label.
Described reference element includes but are not limited to down one or more of column element: time, place.
Described reference data includes but are not limited to down one or more of column data: creation-time, nearest modification time.
The described method that label is sorted is:
1) with described tally set T
μ 'And T ' merging, be designated as tally set T
r
2) from T
rIn arbitrary tag element t, it is at described file set P
μ 'The middle number of times that occurs is designated as n
1, the number of times that occurs in described file set Q is designated as n
2
3) according to formula p=β n
1+ vn
2Calculate the possibility value that this tag element occurs in current file, wherein β is to described T
μ 'The weights of middle label, v is the weights to label among the described Q;
4) m label that possibility value rank is forward as a result of returns to the user.
Described weights β, v and described span are by default or be set by the user.
A kind of intelligent prompt method of file label, the file at adding label the steps include:
1) from context, extracts a reference element value;
2) set a span according to the reference element value;
3) extract the file of reference element value in this span, obtain file set P
μ 'And corresponding tally set T
μ '
4) statistics T
μ '-T
KnownIn each element and T
KnownIn each element at file set P
μ 'File in the times N that occurs simultaneously
1, T wherein
KnownFor all had added the tally set of label file;
5) from user interaction history extracting data one reference data value;
6) set a span according to this reference data value;
7) extract the file of reference data value in this span, obtain file set Q and corresponding tally set T ' thereof;
8) statistics T
μ '-T
KnownIn each element and T
KnownIn each element at file set P
μ 'File in the times N that occurs simultaneously
2
9) times N that occurs according to each tag element
1And N
2Label is sorted, and the prompting user selects label.
Described reference element includes but are not limited to down one or more of column element: time, place.
Described reference data includes but are not limited to down one or more of column data: creation-time, nearest modification time.
The described method that label is sorted is:
1) with described tally set T
μ '-T
KnownWith tally set T '-T
KnownMerge, be designated as tally set T
r
2) from T
rIn arbitrary tag element t, it is at described file set P
μ 'The middle number of times that occurs is designated as n
1, the number of times that occurs in described file set Q is designated as n
2
3) according to formula p=β n
1+ vn
2Calculate the possibility value that this tag element occurs in current file, wherein β is to described T
μ 'The weights of middle label, v is the weights to label among the described Q;
4) m label that possibility value rank is forward as a result of returns to the user.
Described weights β, v and described span are by default or be set by the user.
Good effect of the present invention is:
The present invention can carry out effective label prompting to the user, helps the user to add label for file.Compared with prior art, main difference part of the present invention is to have considered simultaneously user's the interactive history and the context of file content, thereby points out for the user provides the label that more accurately enriches.
Description of drawings
Fig. 1 main process figure of the present invention;
Fig. 2 shows also not putting on the file of any label, carries out the algorithm flow that label adds prompting;
Fig. 3 shows indicating the file of some labels, carries out the algorithm flow that label adds prompting.
Embodiment
Add reminding method with introducing label of the present invention in detail below.For the file that some labels are arranged with never added the file of label, we will separately discuss.Method mainly can be divided into following three steps: one, the context of analyzing and processing file content; Two, the historical data of analyzing and processing user interactions; Three, the result of comprehensive first two steps provides call tag.Wherein first and second two steps can walk abreast and carry out, as shown in Figure 1.
Provide detailed introduction below.
The known file set is designated as P, and its corresponding tally set is designated as T, and number of files wherein is n., the file p among the file set P wherein
i(1＜i＜k, k＜, also have n-k file not put on any label n) for having added label.Tally set T is the set that All Files contained among the P label is formed.Be that example illustrates the method applied in the present invention only below with the content context of photo.For the photo that digital camera is taken, its shooting time can directly extract (digital camera that has also provides spot for photography information) from photo files.Thus, obtain the shooting time set of file set P correspondence, promptly the shooting time with all photos among the P is the set of element, is designated as U.If camera can provide the information of spot for photography, shooting time parameter hereinafter can be replaced with location parameters so, carry out the processing of all fours.
To a photo p who does not also put on any label
j(j＜n), its label prompting algorithm flow as shown in Figure 2.The contextual step of analyzing and processing file content is as follows: at first, obtain its shooting time μ, traversal time set U finds shooting time μ ' afterwards, makes μ-μ '＜ω, and wherein ω is a time threshold; Then, find all photos, be designated as photograph collection P in time μ ' shooting
μ ', its corresponding tally set is designated as T
μ 'At last, statistics T
μ 'In each element at P
μ 'The number of times that occurs in all photos.
The step of analyzing and processing user interaction history data is as follows: at first, obtain nearest l time of user's the time of opening this document, the set that these time points are formed is designated as R; Then, each the element α among the R finds out that (set that these files are formed is designated as Q for α-σ, the every other file that the user opens in the time period of α+σ), and its corresponding tally set is designated as T '; At last, the number of times that each element occurs in the All Files in Q among the statistics T '.
After intact context of analyzing and processing and interactive history data, obtain tally set T
μ 'And T ', and know the wherein number of times of each element appearance in corresponding file is gathered separately.Next need from these two tag sets, to select some labels,, return to the user as prompt options.Introduce a kind of screening technique below.At first to tally set T
μ 'And T ' asks union, is designated as T
rTo T
rIn arbitrary element t, suppose that it is at P
μ 'The middle number of times that occurs is n
1, the number of times that occurs in Q is n
2, its possibility value p=β n that in current file, occurs then
1+ vn
2, wherein β is to T
μ 'The weights of middle label, v is the weights to label among the Q.At last, m the label that possibility value rank is forward as a result of returns to the user.
In above-mentioned processing procedure, the label number m that time threshold ω, file are opened number of times l, time span σ, weights β, weights v recently and returned to the user, these parameters both can be by systemic presupposition, also can real-time preference learn, constantly adjust according to user's feedback in the rear end to the user.
To indicating the photo p of some labels
i(i＜n), establishing existing tag set is T
Known, its label prompting algorithm flow as shown in Figure 3.The contextual step of analyzing and processing file content is as follows: at first, obtain its shooting time μ, traversal time set U finds shooting time μ ' afterwards, makes μ-μ '＜ω, and wherein ω is a time threshold; Then, find all photos, be designated as photograph collection P in time μ ' shooting
μ ', its corresponding tally set is designated as T
μ 'At last, statistics T
μ '-T
KnownIn each element and T
KnownIn element at P
μ 'Photo in the number of times that occurs simultaneously, statistical method is as follows: for T
μ '-T
KnownIn each element t, traversal set P
μ ', statistics t and T
KnownMiddle element is at P
μ 'All photos in the frequency n that occurs simultaneously
1
The step of analyzing and processing user interaction history data is as follows: at first, obtain nearest l time of user's the time of opening this document, the set that these time points are formed is designated as R; Then, each the element α among the R finds out that (set that these files are formed is designated as Q for α-σ, the every other file that the user opens in the time period of α+σ), and its corresponding tally set is designated as T '; At last, statistics T '-T
KnownIn each element and T
KnownIn the number of times that in the file of Q, occurs simultaneously of element, statistical method is as follows: for T '-T
KnownIn each element t, traversal set Q, statistics t and T
KnownThe frequency n that middle element occurs in the All Files of Q simultaneously
2
After intact context of analyzing and processing and interactive history data, obtain tally set T
μ '-T
KnownAnd T '-T
Known, and know wherein each element and T
KnownThe number of times that middle element occurs in the corresponding file set.Next need from these two tag sets, to select some labels,, return to the user as prompt options.Introduce a kind of screening technique below.At first to tally set T
μ '-T
KnownAnd T '-T
KnownAsk union, be designated as T
rTo T
rIn arbitrary element t, known its corresponding n
1With n
2, its possibility value p=β n that in current file, occurs then
1+ vn
2, wherein β is to T
μ '-T
KnownThe weights of middle label, v is to T '-T
KnownThe weights of middle label.At last, m the label that possibility value rank is forward as a result of returns to the user.
In above-mentioned processing procedure, the label number m that time threshold ω, file are opened number of times l, time span σ, weights β, weights v recently and returned to the user, these parameters both can be by systemic presupposition, also can real-time preference learn, constantly adjust according to user's feedback in the rear end to the user.
Provided for the file that some labels are arranged above and never added the file of label, carried out the algorithm that label adds prompting.Can see the context of this method by the extraction document content, the interactive history of analysis user and operating system, the possibility value that computation tag occurs in current file, the label that possibility value rank is forward returns to the user as prompt options.And can dynamically adjust each parameter in the method by the machine learning algorithm of rear end.
We have provided the flow process of file label intelligent prompt algorithm hereinbefore, have provided detailed specific embodiments, and all can realize by the mode of software programming, and software programming need not creative work for those of ordinary skills and just can realize.
List of references
[1]G.Bell，The?Cyber?All?Project：A?Personal?Store?for?Everything，MicrosoftResearch?Technical?Report?MSR-2000-75，July?2000.
[2]ACDSee.http://www.acdsee.com
[3]Microsoft?Windows?Photo?Gallery.
http://www.microsoft.com/windows/products/windowsvista/seeit/sharephotos/default.mspx
[4]Adobe?Photoshop?Album.http://www.adobe.com/products/photoshopalbum
[5]Google?Picasa.http://picasa.google.com
[6]Flickr.com.http://www.flickr.com
[7]K.Rodden?and?K.R.Wood，How?Do?People?Manage?Their?Digital?Photographs，CHI’2003.
[8]M.Ames?and?M.Naaman，Why?We?Tag：Motivations?for?Annotation?in?Mobile?andOnline?Media，CHI’2007.
[9]M.Cooper，J.Foote，A.Girgensohn?and?L.Wilcox，Temporal?Event?Clusteringfor?Digital?Photo?Collections，ACM?Transactions?on?Multimedia?Computing，Communications?and?Applications，Vol.1，No.3，Pages?269-288，August?2005.
[10]M.Naaman，Y.J.Song，A.Paepcke?and?H.G-Molina，Automatic?Organization?forDigital?Photographs?with?Geographic?Coordinates，JCDL’04.
[11]T.Mei，B.Wang，X-S.Hua，H-Q?Zhou?and?S.Li，Probabilistic?MultimodalityFusion?for?Event?Based?Home?Photo?Clustering，ICME’2006.
[12]J.Cui，F.Wen，R.Xiao，Y.Tian?and?X.Tang，EasyAlbum：AnInteractive?PhotoAnnotation?System?Based?on?Face?Clustering?and?Re-ranking，CHI’2007.
[13]S.Yang?and?Y.M.Ro，Photo?Indexing?Using?Person-based?Multi-feature?Fusionwith?Temporal?Context，International?Conference?on?Mobile?Ubiquitous?Computing，Systems，Services?and?Technologies，2007.
[14]M.Naaman，R.B.Yeh，H.G-Molina?and?A.Paepcke，Leveraging?Context?to?ResolveIdentityin?Photo?Albums，JCDL’05.
[15]T.Blanc-Brude?and?D.L.Scapin.What?Do?People?Recall?about?Their?Documents？Implications?for?Desktop?Search?Tools，IUI’07，pages?102-111，New?York，NY，USA，2007.
[16]K.Gyllstrom?and?C.Soules.Seeing?Is?Retrieving：Building?Information?Contextfrom?What?The?User?Sees，IUI’08，pages?189-198，Maspalomas，Gran?Canaria，Spain，2008.
Claims (6)
1. the intelligent prompt method of a file label, the file at not adding label the steps include:
1) from context, extracts a reference element value; Described reference element is: creation-time or nearest modification time;
2) set a span according to the reference element value;
3) extract the file of reference element value in this span, obtain file set P
μ 'And corresponding tally set T
μ '
4) statistics tally set T
μ 'In each element at file set P
μ 'The middle times N that occurs
1
5) from the nearest time of opening current file l time of user interaction history extracting data, be designated as set R;
6) each element among the pair set R is set a span;
7) be extracted in the every other file that the user opens in this span, obtain file set Q and corresponding tally set T ' thereof;
8) times N that each element occurs in file set Q among the statistics tally set T '
2
9) times N that occurs according to each tag element
1And N
2Label is sorted, and the prompting user selects label.
2. the method for claim 1 is characterized in that the described method that label is sorted is:
1) with described tally set T
μ 'And T ' merging, be designated as tally set T
r
2) from T
rIn arbitrary tag element t, it is at described file set P
μ 'The middle number of times that occurs is designated as n
1, the number of times that occurs in described file set Q is designated as n
2
3) according to formula p=β n
1+ ν n
2Calculate the possibility value that this tag element occurs in current file, wherein β is to described T
μ 'The weights of middle label, ν is the weights to label among the described Q;
4) m label that possibility value rank is forward as a result of returns to the user.
3. method as claimed in claim 2 is characterized in that described weights β, ν and described span are by default or be set by the user.
4. the intelligent prompt method of a file label, the file at adding label the steps include:
1) from context, extracts a reference element value; Described reference element is: creation-time or nearest modification time;
2) set a span according to the reference element value;
3) extract the file of reference element value in this span, obtain file set P
μ 'And corresponding tally set T
μ '
4) statistics T
μ '-T
KnownIn each element and T
KnownIn each element at file set P
μ 'File in the times N that occurs simultaneously
1, T wherein
KnownFor all had added the tally set of label file;
5) from the nearest time of opening current file l time of user interaction history extracting data, be designated as set R;
6) each element among the pair set R is set a span;
7) be extracted in the every other file that the user opens in this span, obtain file set Q and corresponding tally set T ' thereof;
8) statistics T
μ '-T
KnownIn each element and T
KnownIn each element at file set P
μ 'File in the times N that occurs simultaneously
2
9) times N that occurs according to each tag element
1And N
2Label is sorted, and the prompting user selects label.
5. method as claimed in claim 4 is characterized in that the described method that label is sorted is:
1) with described tally set T
μ '-T
KnownWith tally set T '-T
KnownMerge, be designated as tally set T
r
2) from T
rIn arbitrary tag element t, it is at described file set P
μ 'The middle number of times that occurs is designated as n
1, the number of times that occurs in described file set Q is designated as n
2
3) according to formula p=β n
1+ ν n
2Calculate the possibility value that this tag element occurs in current file, wherein β is to described T
μ 'The weights of middle label, ν is the weights to label among the described Q;
4) m label that possibility value rank is forward as a result of returns to the user.
6. method as claimed in claim 5 is characterized in that described weights β, ν and described span are by default or be set by the user.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2008101061577A CN100573532C (en) | 2008-05-09 | 2008-05-09 | A kind of intelligent prompt method of file label |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2008101061577A CN100573532C (en) | 2008-05-09 | 2008-05-09 | A kind of intelligent prompt method of file label |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101286174A CN101286174A (en) | 2008-10-15 |
| CN100573532C true CN100573532C (en) | 2009-12-23 |
Family
ID=40058377
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2008101061577A Expired - Fee Related CN100573532C (en) | 2008-05-09 | 2008-05-09 | A kind of intelligent prompt method of file label |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100573532C (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102238268B (en) | 2010-04-30 | 2013-10-30 | 腾讯科技（深圳）有限公司 | Information prompting method and device |
| CN102262638A (en) * | 2010-05-31 | 2011-11-30 | 国际商业机器公司 | Method and device for indexing picture |
| CN102279887B (en) * | 2011-08-18 | 2016-06-01 | 北京百度网讯科技有限公司 | A kind of Document Classification Method, Apparatus and system |
| CN102880687B (en) * | 2012-09-14 | 2015-07-29 | 北京大学 | Based on individual interactive data retrieval method and the system thereof of label technique |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1746884A (en) * | 2004-09-06 | 2006-03-15 | 英保达股份有限公司 | Automatic naming method and system for digital image file |
| CN1846209A (en) * | 2003-09-04 | 2006-10-11 | 诺基亚有限公司 | Method and arrangement for naming pictures to be saved in a mobile station |
-
2008
- 2008-05-09 CN CNB2008101061577A patent/CN100573532C/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1846209A (en) * | 2003-09-04 | 2006-10-11 | 诺基亚有限公司 | Method and arrangement for naming pictures to be saved in a mobile station |
| CN1746884A (en) * | 2004-09-06 | 2006-03-15 | 英保达股份有限公司 | Automatic naming method and system for digital image file |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101286174A (en) | 2008-10-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109189991B (en) | Duplicate video identification method, device, terminal and computer readable storage medium | |
| US8027541B2 (en) | Image organization based on image content | |
| US9798741B2 (en) | Interactive image selection method | |
| US20120114257A1 (en) | Interactive image selection method | |
| Rudinac et al. | Learning crowdsourced user preferences for visual summarization of image collections | |
| Lee et al. | MAP-based image tag recommendation using a visual folksonomy | |
| Wang et al. | CLARE: A joint approach to label classification and tag recommendation | |
| JP2008123486A (en) | Method, system and program for detecting one or plurality of concepts by digital media | |
| CN100573532C (en) | A kind of intelligent prompt method of file label | |
| Liu et al. | Composing semantic collage for image retargeting | |
| JP4878591B2 (en) | Content search device, content search method, program, and recording medium | |
| CN100462978C (en) | Image searching method and system | |
| CN101286175B (en) | File presentation method and its system based on label | |
| Min et al. | Multimodal spatio-temporal theme modeling for landmark analysis | |
| Tseng et al. | Web image annotation by fusing visual features and textual information | |
| Nguyen et al. | Social tagging analytics for processing unlabeled resources: A case study on non-geotagged photos | |
| Li et al. | Visualization of photo album: selecting a representative photo of a specific event | |
| Firmino et al. | Automatic and semi-automatic annotation of people in photography using shared events | |
| Xie et al. | The accuracy and value of machine-generated image tags: design and user evaluation of an end-to-end image tagging system | |
| Sirirattanapol et al. | Deep image retrieval applied on kotenseki ancient japanese literature | |
| Sun et al. | Photo 4W: Mobile photo management on what, where, who and when | |
| Rabbath et al. | Multimedia retrieval in social networks for photo book creation | |
| Chu et al. | Travelmedia: An intelligent management system for media captured in travel | |
| Dong et al. | Research on a Pedestrian Detection Algorithm Based on Improved SSD Network | |
| Duan et al. | Automatic Re-Ranking in Clip Matching of Stereoscopic Video based on Clustering and Density |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20091223 Termination date: 20170509 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |