You Are What You Tag Yi-Ching Huang and Chia-Chuan Hung and Jane Yung-jen Hsu Department of Computer Science and Information Engineering Graduate Institute.

Slides:



Advertisements
Similar presentations
A Comparison of Implicit and Explicit Links for Web Page Classification Dou Shen 1 Jian-Tao Sun 2 Qiang Yang 1 Zheng Chen 2 1 Department of Computer Science.
Advertisements

The Usage of Social Networks In Educational Context Sacide Güzin MAZMAN, Yasemin KOÇAK USLUEL Hacettepe University, Faculty of Education Department of.
Prediction Modeling for Personalization & Recommender Systems Bamshad Mobasher DePaul University Bamshad Mobasher DePaul University.
A Graph-based Recommender System Zan Huang, Wingyan Chung, Thian-Huat Ong, Hsinchun Chen Artificial Intelligence Lab The University of Arizona 07/15/2002.
Introduction Information Management systems are designed to retrieve information efficiently. Such systems typically provide an interface in which users.
WebMiningResearch ASurvey Web Mining Research: A Survey Raymond Kosala and Hendrik Blockeel ACM SIGKDD, July 2000 Presented by Shan Huang, 4/24/2007.
Modern Information Retrieval Chapter 2 Modeling. Can keywords be used to represent a document or a query? keywords as query and matching as query processing.
Commentary-based Video Categorization and Concept Discovery By Janice Leung.
Web Mining Research: A Survey
WebMiningResearchASurvey Web Mining Research: A Survey Raymond Kosala and Hendrik Blockeel ACM SIGKDD, July 2000 Presented by Shan Huang, 4/24/2007 Revised.
Recommender systems Ram Akella February 23, 2011 Lecture 6b, i290 & 280I University of California at Berkeley Silicon Valley Center/SC.
1 Information Retrieval and Extraction 資訊檢索與擷取 Chia-Hui Chang, Assistant Professor Dept. of Computer Science & Information Engineering National Central.
Presented by Zeehasham Rasheed
TextMOLE: Text Mining Operations Library and Environment Daniel B. Waegel and April Kontostathis, Ph.D. Ursinus College Collegeville PA.
Recommender systems Ram Akella November 26 th 2008.
Center for E-Business Technology Seoul National University Seoul, Korea Social Network Collaborative Filtering Research Meeting Babar Tareen
CHAMELEON : A Hierarchical Clustering Algorithm Using Dynamic Modeling
Using Friendship Ties and Family Circles for Link Prediction Elena Zheleva, Lise Getoor, Jennifer Golbeck, Ugur Kuter (SNAKDD 2008)
Web 2.0: Concepts and Applications 4 Organizing Information.
Tag Clouds Revisited Date : 2011/12/12 Source : CIKM’11 Speaker : I- Chih Chiu Advisor : Dr. Koh. Jia-ling 1.
JASS 2005 Next-Generation User-Centered Information Management Information visualization Alexander S. Babaev Faculty of Applied Mathematics.
PAUL ALEXANDRU CHIRITA STEFANIA COSTACHE SIEGFRIED HANDSCHUH WOLFGANG NEJDL 1* L3S RESEARCH CENTER 2* NATIONAL UNIVERSITY OF IRELAND PROCEEDINGS OF THE.
UOS 1 Ontology Based Personalized Search Zhang Tao The University of Seoul.
Ontology-Driven Automatic Entity Disambiguation in Unstructured Text Jed Hassell.
Data Mining Chapter 1 Introduction -- Basic Data Mining Tasks -- Related Concepts -- Data Mining Techniques.
Chengjie Sun,Lei Lin, Yuan Chen, Bingquan Liu Harbin Institute of Technology School of Computer Science and Technology 1 19/11/ :09 PM.
TOPIC CENTRIC QUERY ROUTING Research Methods (CS689) 11/21/00 By Anupam Khanal.
ON THE SELECTION OF TAGS FOR TAG CLOUDS (WSDM11) Advisor: Dr. Koh. Jia-Ling Speaker: Chiang, Guang-ting Date:2011/06/20 1.
Developing Trust Networks based on User Tagging Information for Recommendation Making Touhid Bhuiyan et al. WISE May 2012 SNU IDB Lab. Hyunwoo Kim.
Information Visualization: Ten Years in Review Xia Lin Drexel University.
Attributed Visualization of Collaborative Workspaces Mao Lin Huang, Quang Vinh Nguyen and Tom Hintz Faculty of Information Technology University of Technology,
CONCLUSION & FUTURE WORK Normally, users perform search tasks using multiple applications in concert: a search engine interface presents lists of potentially.
You Are What You Tag Yi-Ching (Janet) Huang Chia-Chuan (Evelyn) Hung Jane Yung-jen Hsu From National Taiwan University 2008/03/26.
Personalized Course Navigation Based on Grey Relational Analysis Han-Ming Lee, Chi-Chun Huang, Tzu- Ting Kao (Dept. of Computer Science and Information.
Semantic Wordfication of Document Collections Presenter: Yingyu Wu.
2005/12/021 Content-Based Image Retrieval Using Grey Relational Analysis Dept. of Computer Engineering Tatung University Presenter: Tienwei Tsai ( 蔡殿偉.
2005/12/021 Fast Image Retrieval Using Low Frequency DCT Coefficients Dept. of Computer Engineering Tatung University Presenter: Yo-Ping Huang ( 黃有評 )
Intelligent Database Systems Lab N.Y.U.S.T. I. M. Externally growing self-organizing maps and its application to database visualization and exploration.
Department of Information Science and Applications Hsien-Jung Wu 、 Shih-Chieh Huang Asia University, Taiwan An Intelligent E-learning system for Improving.
WEB 2.0 PATTERNS Carolina Marin. Content  Introduction  The Participation-Collaboration Pattern  The Collaborative Tagging Pattern.
Digital Libraries Lillian N. Cassel Spring A digital library An informal definition of a digital library is a managed collection of information,
Harvesting Social Knowledge from Folksonomies Harris Wu, Mohammad Zubair, Kurt Maly, Harvesting social knowledge from folksonomies, Proceedings of the.
From Text to Image: Generating Visual Query for Image Retrieval Wen-Cheng Lin, Yih-Chen Chang and Hsin-Hsi Chen Department of Computer Science and Information.
Digital Libraries1 David Rashty. Digital Libraries2 “A library is an arsenal of liberty” Anonymous.
A Novel Visualization Model for Web Search Results Nguyen T, and Zhang J IEEE Transactions on Visualization and Computer Graphics PAWS Meeting Presented.
Personal Tag Semantic Relation Yi-Ching Huang 2008/02/27 Yi-Ching Huang 2008/02/27.
Commonsense Reasoning in and over Natural Language Hugo Liu, Push Singh Media Laboratory of MIT The 8 th International Conference on Knowledge- Based Intelligent.
1 Adaptive Subjective Triggers for Opinionated Document Retrieval (WSDM 09’) Kazuhiro Seki, Kuniaki Uehara Date: 11/02/09 Speaker: Hsu, Yu-Wen Advisor:
User Modeling and Recommender Systems: recommendation algorithms
Hybrid Content and Tag-based Profiles for recommendation in Collaborative Tagging Systems Latin American Web Conference IEEE Computer Society, 2008 Presenter:
MMM2005The Chinese University of Hong Kong MMM2005 The Chinese University of Hong Kong 1 Video Summarization Using Mutual Reinforcement Principle and Shot.
Relevance Feedback in Image Retrieval System: A Survey Tao Huang Lin Luo Chengcui Zhang.
Using category-Based Adherence to Cluster Market-Basket Data Author : Ching-Huang Yun, Kun-Ta Chuang, Ming-Syan Chen Graduate : Chien-Ming Hsiao.
Instance Discovery and Schema Matching With Applications to Biological Deep Web Data Integration Tantan Liu, Fan Wang, Gagan Agrawal {liut, wangfa,
CiteData: A New Multi-Faceted Dataset for Evaluating Personalized Search Performance CIKM’10 Advisor : Jia-Ling, Koh Speaker : Po-Hsien, Shih.
INFORMATION RETRIEVAL MEASUREMENT OF RELEVANCE EFFECTIVENESS 1Adrienn Skrop.
A Self-organizing Semantic Map for Information Retrieval Xia Lin, Dagobert Soergel, Gary Marchionini presented by Yi-Ting.
GUILLOU Frederic. Outline Introduction Motivations The basic recommendation system First phase : semantic similarities Second phase : communities Application.
InterestMap - Harvesting Social Network Profiles for Recommendation Hugo Liu (MIT Media lab) Pattie Maes (MIT Media lab) Speaker: Huang, Yi-Ching.
Introduction to Machine Learning, its potential usage in network area,
Designing Cross-Language Information Retrieval System using various Techniques of Query Expansion and Indexing for Improved Performance  Hello everyone,
Introduction Multimedia initial focus
Personalized Social Image Recommendation
Using Friendship Ties and Family Circles for Link Prediction
Multimedia Information Retrieval
Ying Dai Faculty of software and information science,
Ying Dai Faculty of software and information science,
Ying Dai Faculty of software and information science,
WSExpress: A QoS-Aware Search Engine for Web Services
Presentation transcript:

You Are What You Tag Yi-Ching Huang and Chia-Chuan Hung and Jane Yung-jen Hsu Department of Computer Science and Information Engineering Graduate Institute of Networking and Multimedia National Taiwan University Association for the Advancement of Artificial Intelligence(AAAI ), 2008 Presenter: Hsiao-Pei, Chang

Outline Introduction Related Works Tags Analysis and Personal Profile Semantic Similarity and Co-occurrence Visualization Conclusions and Comments 2010/01/082Speaker: Hsiao-Pei,Chang

Introduction People establish personal profiles to obtain online services. Profiles consisting of simple factual data provide an inadequate description of the individual, as they are often incomplete, mostly subjective and can not reflect dynamic changes. 2010/01/083Speaker: Hsiao-Pei,Chang

Introduction This paper presents an novel idea of “you are what you tag” for user profiling. An individual can be effectively profiled by the tags associated with his/her social media. The basic assumption is that the rich online media produced/consumed by an individual can reveal important features about the person. 2010/01/084Speaker: Hsiao-Pei,Chang

Introduction Their research explores tag-based user profiling on several levels. This paper presents the personal, social, and global views of an individual’s profile based on the tags and content of his/her social bookmarks. 2010/01/085Speaker: Hsiao-Pei,Chang

Introduction When many people tag a person’s content collection with a specific keyword, it is natural to assume that the content owner shares the same idea. To facilitate the alternative views, profiles are visualized as tag clouds based on color harmonic combinations. Commonsense semantic analysis and co- occurrence measurement are defined to calculate tag similarity. 2010/01/086Speaker: Hsiao-Pei,Chang

Related Work InterestMap ◦ Research on InterestMap (Liu & Maes 2005) harvests profiles from social networking websites, ex.MySpace ◦ Rather than traditional recommender systems, they recommend by considering the interest of people instead of the historical behavior in a particular application. ◦ This work not only upgrade the accuracy of recommendation but also provide a visual way to explore one’s interests on InterestMap. 2010/01/087Speaker: Hsiao-Pei,Chang

Related Work 2010/01/088Speaker: Hsiao-Pei,Chang

Related Work Tagging is a social indexing process and contents can be categorized by any number of tags. The tag cloud is a visual interface to help people retrieve important information quickly. In (Kaser & Lemire 2007), some models and algorithms to improve the display of tag cloud in HTML were presented. 2010/01/089Speaker: Hsiao-Pei,Chang

Related Work Different from InterestMap, this paper utilize the tags in these social media content which people collected. All tags are grouped by their concepts for visual layout rather than the alphabetical order in a traditional tag cloud. 2010/01/0810Speaker: Hsiao-Pei,Chang

Related Work This paper measure three different viewpoints to provide a comprehensive representation of a person. People can then compare these three profile models using our visualization function to learn more information about a person. 2010/01/0811Speaker: Hsiao-Pei,Chang

Tag and Social Network In this paper, they propose using tags to profile a person instead of a typical profile list. And for simplicity, social bookmarks are used as the data source, and each bookmarked document is assumed to have multiple tags. 2010/01/0812Speaker: Hsiao-Pei,Chang

Tag and Social Network The advantages of using tags include: ◦ the flexibility in describing a person using any term ◦ the precision in presenting a person’s preferences. ◦ the ease in showing the different views of a person. This paper assume that the social network is given. 2010/01/0813Speaker: Hsiao-Pei,Chang

Data Modeling This paper propose to add social media as a new type of nodes into social network. We call this network a Social Media Network. 2010/01/0814Speaker: Hsiao-Pei,Chang

Data Modeling For each arrow, the floating number denotes the capacity of this tag to describe this document..Each tag has the strength attribute to represent the importance of the tag for its owner. 2010/01/0815Speaker: Hsiao-Pei,Chang

Tag Analysis This paper define a value, capacity, to represent how much a tag can describe the content of a document. The importance of a tag to an owner, denoted as tag strength, is determined by analyzing its capacity, the volume of covered bookmarks, and the quality of its bookmarked content. 2010/01/0816Speaker: Hsiao-Pei,Chang

Tag Capacity A tag used by more people to describe an identical bookmark link has a higher importance to this bookmarked document. The first tag may be more relevant than the second tag when a user tags this document. Thus they consider tagging frequency, tagging order, and tags diversity when trying to determine the capacity of a tag to describe a bookmarked document. 2010/01/0817Speaker: Hsiao-Pei,Chang

Tag Capacity A collection of bookmarked documents as D,and a set of tags T which are tagged on D. b = (p, bT, d) which means a person p who tagged document d ∈ D with a sequence of tags bT={bt 1, bt 2,……, bt n }. They first define the order weight of tag bt i ∈ bT as 2010/01/0818Speaker: Hsiao-Pei,Chang

Tag Capacity They use ConceptNet and co-occurrence measurements to group tags in bT with their concept. Let bT c ={bt 1, bt 2,……, bt k } be a cluster of bT with k tags, then we define the concept weight of ∈ bT c as 2010/01/0819Speaker: Hsiao-Pei,Chang

Tag Capacity Combining both weight measurements, for each bt i ∈ bT,we can define the weight of bt i for a document d given by a person p as 2010/01/0820Speaker: Hsiao-Pei,Chang

Tag Capacity we can determine the importance of a tag t for a document d. M as a set of people who ever tagged document d with tag t, thus the capability of t on d is 2010/01/0821Speaker: Hsiao-Pei,Chang

Personal Profile Each tag has different importance in representing a person. And each tag has a different capacity in describing a document. So they summarize the weights of an identical tag on one’s own documents, and obtain a total weight of a tag for representing this person. This paper define three types of viewpoints to consider one’s bookmarks: personal, social, and global views to obtain more objective facts. 2010/01/0822Speaker: Hsiao-Pei,Chang

Personal Profile Personal viewpoint ◦ Only consider the tags assigned by p, and denote these candidates as a set CT p. ◦ For each candidate tag t ∈ CT p, we define the strength of t from p’s view as ◦ Where sub index p means from p’s view. Thus ◦ They set a threshold to filter those tags with strength lower than threshold. 2010/01/0823Speaker: Hsiao-Pei,Chang

Personal Profile Social viewpoint ◦ Consider the tags assigned by the actors on p’s personal social network. The candidate tags are those tags which were assigned by p’s acquaintances on these documents. 2010/01/0824Speaker: Hsiao-Pei,Chang

Personal Profile Global viewpoint ◦ Consider the tags assigned by everyone on D p. This viewpoint can be thought of as a generalization of the social viewpoint. ◦ The results of this viewpoint can reflect the common opinions from public, and probably is most objective. 2010/01/0825Speaker: Hsiao-Pei,Chang

Semantic Similarity and Co-occurrence To cluster the tags automatically, they propose two methods for calculating tag similarity and cluster the ones with high similarity. ◦ commonsense reasoning similarity ◦ relative co-occurrence statistics. 2010/01/0826Speaker: Hsiao-Pei,Chang

The ConceptNet-based semantic similarity The ConceptNet-based semantic similarity This paper use commonsense reasoning to obtain related tags with a similar concept. ConceptNet is a freely available commonsense knowledgebase and it provides a natural-language-processing toolkit for reasoning tasks including “topic-jisting”, “affect-sensing ”, “ analogy- making”, and “text summarization”. 2010/01/0827Speaker: Hsiao-Pei,Chang

The ConceptNet-based semantic similarity The ConceptNet-based semantic similarity The ConceptNet toolkit provides node- level and document-level reasoning operations. 2010/01/0828Speaker: Hsiao-Pei,Chang

The ConceptNet-based semantic similarity The ConceptNet-based semantic similarity Two functions on textual analysis(Liu& Singh 2004) are introduced: ◦ GetContext():  It accepts the input of a textual document which is then translated into a ConceptNet-compatible format.  It finds the neighboring relevant concepts using spreading activation around this concept of document. ◦ GuessConcept():  It outputs a list of potential items which are analogous to the input concept. 2010/01/0829Speaker: Hsiao-Pei,Chang

The ConceptNet-based semantic similarity The ConceptNet-based semantic similarity In this paper, GuessConcept() use to acquire a list of analogous tags given by a tag. Then, they use GetContext() to acquire all neighboring relevant tags given by analogous tags using spreading activation. The energy of any tag after a spreading step is calculated : 2010/01/0830Speaker: Hsiao-Pei,Chang

Co-occurrence They propose an co-occurrence measurement to compute the relationship between two different and irrelevant tags. Semantic Similarity Between Tags: Given two tags a and b, we could obtain two sets C a and C b including the neighboring relevant tags of a and b, respectively. 2010/01/0831Speaker: Hsiao-Pei,Chang

Visualization To facilitate the alternative views, profiles are visualized as tag clouds based on color harmonic combinations. They design a way of visualization, that is, by grouping tags with similar concepts and displaying them with similar colors, viewers can be more attentive to focus on the information he/she needs. 2010/01/0832Speaker: Hsiao-Pei,Chang

Visualization 2010/01/0833Speaker: Hsiao-Pei,Chang

Personal viewpoint 2010/01/0834Speaker: Hsiao-Pei,Chang

Social viewpoint 2010/01/0835Speaker: Hsiao-Pei,Chang

Global viewpoint 2010/01/0836Speaker: Hsiao-Pei,Chang

Conclusions This paper presented an novel approach to user profiling based on the tags associated with one’s social media. Three alternative views, personal, social, and global views, are defined to offer a comprehensive profile of an individual. A grouped tag cloud to visualize a person’s profile, viewers can easily focus on the most important and relevant information. 2010/01/0837Speaker: Hsiao-Pei,Chang

Comments The interests or characteristics of people change over time. A static picture is unable to represent temporal flow. Whether this approach can help to learn English or not ? 2010/01/0838Speaker: Hsiao-Pei,Chang