recommender systems urank
play

Recommender Systems + uRank Multimedia Information Systems 2 - PowerPoint PPT Presentation

Recommender Systems + uRank Multimedia Information Systems 2 Cecilia di Sciascio cdisciascio@know-center.at KTI, TU Graz April 19 th , 2016 MMIS2 Recommendation Visualization April 19th, 2016 KTI/TU Graz, Know-Center Recommender


  1. Recommender Systems + uRank Multimedia Information Systems 2 Cecilia di Sciascio cdisciascio@know-center.at KTI, TU Graz April 19 th , 2016 MMIS2 – Recommendation Visualization April 19th, 2016 KTI/TU Graz, Know-Center

  2. Recommender Interfaces MMIS2 – Recommendation Visualisation April 19th, 2016 2 KTI/TU Graz, Know-Center

  3. Motivation Engage the user to visit recommended items Swearingen and Sinha (2001). “Beyond Algorithms: An HCI Perspective on Recommender Systems”. MMIS2 – Recommendation Visualisation April 19th, 2016 3 KTI/TU Graz, Know-Center

  4. Recommender Systems MMIS2 – Recommendation Visualisation April 19th, 2016 4 KTI/TU Graz, Know-Center

  5. Recommender Systems MMIS2 – Recommendation Visualisation April 19th, 2016 5 KTI/TU Graz, Know-Center

  6. Recommender Systems MMIS2 – Recommendation Visualisation April 19th, 2016 6 KTI/TU Graz, Know-Center

  7. Recommender systems MMIS2 – Recommendation Visualisation April 19th, 2016 7 KTI/TU Graz, Know-Center

  8. Recommender systems MMIS2 – Recommendation Visualisation April 19th, 2016 8 KTI/TU Graz, Know-Center

  9. Recommender Interfaces What do users expect? • easy to access information about recommended item, • easy to generate new recommendations • classify results by degree of confidence (e.g., best, worst) • understand what is driving recommendations • interface: more like this, new items… MMIS2 – Recommendation Visualisation April 19th, 2016 9 KTI/TU Graz, Know-Center

  10. Recommender Interfaces Gretarsson et al. SmallWorlds: Visualizing Social Recommendations. MMIS2 – Recommendation Visualisation April 19th, 2016 10 KTI/TU Graz, Know-Center

  11. Recommender Interfaces Devendorf , O’Donovan, Höllerer TopicLens: An Interactive Recommender System based on Topical and Social Connections. MMIS2 – Recommendation Visualisation April 19th, 2016 11 KTI/TU Graz, Know-Center

  12. Recommender Interfaces Bostandjiev , O’Donovan, Hollerer. LinkedVis: Exploring Social and Semantic Career Recommendations. MMIS2 – Recommendation Visualisation April 19th, 2016 12 KTI/TU Graz, Know-Center

  13. uRank MMIS2 – Recommendation Visualisation April 19th, 2016 13 KTI/TU Graz, Know-Center

  14. EEXCESS RS plugin MMIS2 – Recommendation Visualisation April 19th, 2016 14 KTI/TU Graz, Know-Center

  15. Why uRank? MMIS2 – Recommendation Visualisation April 19th, 2016 15 KTI/TU Graz, Know-Center

  16. Why uRank? Transparency Controllability + MMIS2 – Recommendation Visualisation April 19th, 2016 16 KTI/TU Graz, Know-Center

  17. uRank UI Web-based tool for document collection exploration MMIS2 – Recommendation Visualisation April 19th, 2016 17 KTI/TU Graz, Know-Center

  18. uRank UI Tag Box Collection Keywords Overview MMIS2 – Recommendation Visualisation April 19th, 2016 18 KTI/TU Graz, Know-Center

  19. uRank UI List + Ranking View Tag Box Ranked list + document similarity & term contribution Collection Keywords Overview MMIS2 – Recommendation Visualisation April 19th, 2016 19 KTI/TU Graz, Know-Center

  20. uRank UI Query Box List + Ranking View Tag Box Ranked list + document similarity & term contribution Collection Keywords Overview MMIS2 – Recommendation Visualisation April 19th, 2016 20 KTI/TU Graz, Know-Center

  21. uRank UI Doc. Viewer Detailed view of document metadata + snippet MMIS2 – Recommendation Visualisation April 19th, 2016 21 KTI/TU Graz, Know-Center

  22. Keyword Extraction Tokenization “women” Document Vectors woman: 3.5 gender: 5.2 wage: 2.1 gap: 1.8 POS Tagging “women” is a noun Singularizatio Collection Keywords n & Lower- • Sorted by document frequency “women”  “woman” casing • Representative term selected Stemming “woman”  “woman” TF-IDF “woman” = 3.5 in d i MMIS2 – Recommendation Visualisation April 19th, 2016 22 KTI/TU Graz, Know-Center

  23. Ranking Method Rank by overall score Score for term t in document d Overall score of document d MMIS2 – Recommendation Visualisation April 19th, 2016 23 KTI/TU Graz, Know-Center

  24. uRank Code Structure TagCloud Urank-Loader TagBox Urank ContentList DocViewer KeywordExtractor RankingModel VisCanvas ranking RankingArray … MMIS2 – Recommendation Visualisation April 19th, 2016 24 KTI/TU Graz, Know-Center

  25. uRank Code Structure TagCloud entry point Urank-Loader TagBox instance Urank ContentList DocViewer KeywordExtractor RankingModel VisCanvas ranking RankingArray … MMIS2 – Recommendation Visualisation April 19th, 2016 25 KTI/TU Graz, Know-Center

  26. uRank Code Structure TagCloud Urank-Loader Core TagBox Urank ContentList DocViewer KeywordExtractor RankingModel VisCanvas ranking RankingArray … MMIS2 – Recommendation Visualisation April 19th, 2016 26 KTI/TU Graz, Know-Center

  27. uRank Code Structure TagCloud Urank-Loader TagBox Urank ContentList DocViewer KeywordExtractor RankingModel VisCanvas ranking RankingArray … Model MMIS2 – Recommendation Visualisation April 19th, 2016 27 KTI/TU Graz, Know-Center

  28. uRank Code Structure Blocks TagCloud Urank-Loader TagBox Urank ContentList DocViewer KeywordExtractor RankingModel VisCanvas ranking RankingArray … MMIS2 – Recommendation Visualisation April 19th, 2016 28 KTI/TU Graz, Know-Center

  29. uRank Code Structure TagCloud Urank-Loader TagBox Urank ContentList DocViewer KeywordExtractor RankingModel VisCanvas ranking RankingArray your vis?? Views MMIS2 – Recommendation Visualisation April 19th, 2016 29 KTI/TU Graz, Know-Center

  30. uRank Code Structure TagCloud Urank TagBox UrankController ContentList DocViewer KeywordExtractor RankingModel VisCanvas ranking RankingArray your vis? Vis MMIS2 – Recommendation Visualisation April 19th, 2016 30 KTI/TU Graz, Know-Center

  31. uRank API Libraries • jQuery: core • D3: stacked bars rendering • jQuery-ui: drag & drop feature • Posjs + naturalJS: keyword extraction • underscore: operations with arrays • modernizr: file loader MMIS2 – Recommendation Visualisation April 19th, 2016 31 KTI/TU Graz, Know-Center

  32. uRank Controller-Block Interaction In urank.js (controller) Block options specify DOM var options = { selector for root element and contentList: { callback functions root: s.contentListRoot, onItemClicked: EVTHANDLER.onItemClicked, onItemMouseEnter: EVTHANDLER.onItemMouseEnter, onItemMouseLeave: EVTHANDLER.onItemMouseLeave, onFaviconClicked: EVTHANDLER.onFaviconClicked, onWatchiconClicked: EVTHANDLER.onWatchiconClicked, onScroll: EVTHANDLER.onParallelBlockScrolled }, contentList object ... initialization contentList = new ContentList (options.contentList); MMIS2 – Recommendation Visualisation April 19th, 2016 32 KTI/TU Graz, Know-Center

  33. uRank Controller-Block Interaction Constructor In contentList.js function ContentList (arguments) { s = $.extend({ root: '', onItemClicked : function(document){}, Object “s” extends default onItemMouseEnter: function(document){}, settings with options passed onItemMouseLeave: function(document){}, by urank.js onFaviconClicked: function(document){}, onWatchiconClicked: function(document){} }, arguments); } ... Events are bound to event handlers $(li).data('id', id) in “s” .on({ click: function(event){ s.onItemClicked .call(this, li.data('id')); }, ... MMIS2 – Recommendation Visualisation April 19th, 2016 33 KTI/TU Graz, Know-Center

  34. uRank Controller-Block Interaction In contentList.js ContentList.prototype = { Controller (urank.js) can call build: _build, any block method in its reset: _reset, prototype update: _update, hover: _hover, unhover: _unhover, selectListItem: _selectListItem, deselectAllListItems: _deselectAllListItems, highlightListItems: _highlightListItems, clearAllFavicons: _clearAllFavicons, toggleFavicon: _toggleFavicon, toggleWatchListItem: _toggleWatchListItem, clearEffects: _clearEffects, ... }; MMIS2 – Recommendation Visualisation April 19th, 2016 34 KTI/TU Graz, Know-Center

  35. uRank Controller-Block Interaction Back in urank.js This is the method passed as callback for list item click event. EVTHANDLER.onItemClicked : function(documentId) { … contentList.selectListItem (documentId); … }; The method for selecting a list item is actually invoked inside the callback function MMIS2 – Recommendation Visualisation April 19th, 2016 35 KTI/TU Graz, Know-Center

  36. uRank UI How to use it Create the DOM elements that will be containers for the 4 blocks (except Docviewer). QueryBox ContentList RankingView TagBox MMIS2 – Recommendation Visualisation April 19th, 2016 36 KTI/TU Graz, Know-Center

  37. uRank API How to use it Include in html header…  jquery (version 1.10.2 or higher)  modernizr (urank/dependencies/modernizr.js)  urank entry point (urank/urank-loader.js) <script type="text/javascript" src="libs/jquery-1.10.2.js"></script> <script type="text/javascript" src="scripts/urank/dependencies/modernizr.js"></script> <script type="text/javascript" src="scripts/urank/urank-loader.js"></script> MMIS2 – Recommendation Visualisation April 19th, 2016 37 KTI/TU Graz, Know-Center

Recommend


More recommend