Your SlideShare is downloading. ×
0
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Identification and Characterization of Events in Social Media
Upcoming SlideShare
Loading in...5
×

Thanks for flagging this SlideShare!

Oops! An error has occurred.

×
Saving this for later? Get the SlideShare app to save on your phone or tablet. Read anywhere, anytime – even offline.
Text the download link to your phone
Standard text messaging rates apply

Identification and Characterization of Events in Social Media

2,411

Published on

Published in: Technology, Business
0 Comments
7 Likes
Statistics
Notes
  • Be the first to comment

No Downloads
Views
Total Views
2,411
On Slideshare
0
From Embeds
0
Number of Embeds
6
Actions
Shares
0
Downloads
68
Comments
0
Likes
7
Embeds 0
No embeds

Report content
Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
No notes for slide
  • The problems we are trying to solve in this thesis
  • Challenges and opportunities
  • What’s been done in the space, very very very briefly, to introduce our known vs. unknown division
  • Explain that we work in real-time (for the most part) and say we divide the space into unknown and know identification scenarios, then mention the type of even we focus on for each. Also briefly mention that as we discuss in the thesis, these are not disjoint
  • Contributions in order, broken down into identification scenarios (more or less).
  • Contributions in order, broken down into identification scenarios (more or less).
  • Contributions in order, broken down into identification scenarios (more or less).
  • Contributions in order, broken down into identification scenarios (more or less).
  • Before we identify trending events, we asked ourselves what types of trending events exist in social media and how are they different from non-event trends that exhibit similar temporal behavior
  • Over 200 million users
  • Give a brief example of each feature type.
  • These will help guide our features for event classification next…
  • Leader-follower?Dropping old clusters, merging clusters, etc. for future work
  • NDCG is a precision-based metric that takes rank into account
  • Bringing it back to point out the parameters
  • This is an outline for the similarity metric learning discussion
  • This is an outline for the similarity metric learning discussion
  • Get the upcoming page for the event with the photo thumbnails , show the machine tag
  • TAGS – ALMOST AS GOOD AS ALL-TEXT
  • LexRank: method for extractive summ. Central nodes are connected to other central nodes, each node has centrality value that it distributes to connected nodes
  • Play with thresholds!
  • … just the ones that went into this thesis 
  • Transcript

    • 1. Identification and Characterization of Events in Social Media<br /> Hila Becker, Thesis Defense<br />
    • 2. Social Media is Changing the World<br />2<br />Lady Gaga, Justin Bieber, and Britney Spears have more Twitter followers than the entire populations of some countries (e.g., Israel, Greece)<br />YouTube is the second largest search engine in the world<br />Every minute, 24 hours of video are uploaded to YouTube<br />Over the past five years people uploaded 6,000,000,000 images to Flickr<br />
    • 3. 3<br />Source: http://www.searchenginejournal.com/the-growth-of-social-media-an-infographic/<br />
    • 4. Event Content in Social Media<br />4<br />
    • 5. 5<br />MIKE CLARKE/AFP/Getty Images<br />
    • 6. 6<br />Source: Tweets from Tahrir, edited by Nadia Idle and Alex Nunns<br />
    • 7. 7<br />
    • 8. Event Identification, Characterization, and Content Selection<br />Identify events and their associated social media documents<br />In a timely manner<br />Across different social media sites<br />Characterize events along different dimensions <br />Select high-quality, relevant, useful event documents<br />8<br />
    • 9. Event Content in Social Media<br />Challenges:<br />Wide variety of topics, not all related to events (e.g., personal status updates, every-day mundane conversations)<br />Unconventional text: abbreviations, typos<br />Large-scale, rapidly produced content<br />Opportunities:<br />Content generated in real-time, as events happen<br />Rich context features (e.g., time, location)<br />Users’ perspective<br />9<br />
    • 10. Event Content in Social Media<br />10<br />Timeliness<br />Real-time<br />Retrospective<br />Twitter new event detection [Petrović et al. NAACL’10]<br />Event detection on Flickr[Chen and Roy CIKM’09]<br />Unknown<br />Content Discovery<br />Organization of YouTube concert videos [Kennedy and Naaman WWW’09]<br />Earthquake prediction using Twitter [Sakaki et al. WWW’10]<br />Known<br />
    • 11. Event Content in Social Media<br />11<br />Trending Event isa real-world occurrence described by:<br />One or more terms and a time period<br />Volume of messages posted for the terms in the time period exceeds some expected level of activity<br />Unknown<br />Content Discovery<br />Planned Event is a real-world occurrence with corresponding published event record consisting of:<br />Title, describing the subject of the event<br />The time at which the event is planned to occur<br />Known<br />
    • 12. Contributions<br />Trend (and trending event) study, for characterizing and differentiating between different types of trends<br />Online clustering framework with an event classification step for identifying trending events and their associated documents in social media<br />Social media document similarity metric learning approaches<br />Query formulation strategies for identifying social media documents for planned events<br />Selection techniques for identifying high quality, relevant, and useful event content<br />12<br />Unknown<br />Known<br />Unknown/Known<br />
    • 13. Contributions<br />Trend (and trending event) study, for characterizing and differentiating between different types of trends<br />Online clustering framework with an event classification step for identifying trending events and their associated documents in social media<br />Social media document similarity metric learning approaches<br />Query formulation strategies for identifying social media documents for planned events<br />Selection techniques for identifying high quality, relevant, and useful event content<br />13<br />Known<br />Unknown/Known<br />
    • 14. Contributions<br />Trend (and trending event) study, for characterizing and differentiating between different types of trends<br />Online clustering framework with an event classification step for identifying trending events and their associated documents in social media<br />Social media document similarity metric learning approaches<br />Query formulation strategies for identifying social media documents for planned events<br />Selection techniques for identifying high quality, relevant, and useful event content<br />14<br />Unknown/Known<br />
    • 15. Contributions<br />Trend (and trending event) study, for characterizing and differentiating between different types of trends<br />Online clustering framework with an event classification step for identifying trending events and their associated documents in social media<br />Social media document similarity metric learning approaches<br />Query formulation strategies for identifying social media documents for planned events<br />Selection techniques for identifying high quality, relevant, and useful event content<br />15<br />
    • 16. Identification and Characterization of Events in Social Media<br />16<br />Characterizationof trending events <br />Identification of trending events <br />Similarity metric learning for trending events<br />Identification of content for planned events<br />Selection of event content<br />
    • 17. What Types of Trends Exist in Social Media?<br />Taxonomy of trends<br />Characterization of each trend<br />Manually assigned categories<br />Automatically computed features<br />Analysis of differences between trend types according to each characteristic<br />17<br />Trending Events<br />Non-Event Trends<br />
    • 18. Trends<br />Trend:<br />One or more terms and a time period <br />Volume of messages posted for the terms in the time period exceeds some expected level of activity<br />May or may not reflect a real-world occurrence<br />A trending event is a type of trend<br />18<br />
    • 19. Twitter Content<br />Streams of textual messages<br />Brief content (140 characters)<br />Communicated to network of followers<br />Provide timely reflection of thoughts and interests<br />19<br />
    • 20. Characterizing Trends on Twitter<br />Collect a set of Twitter trends<br />Burst detection<br />Twitter’s “trending topics”<br />Qualitative analysis: trend taxonomy<br />Quantitative analysis<br />Automatically compute features of each trend and corresponding messages<br />Manually label each trend according to categories introduced by the taxonomy<br />Identify differences between trend categories according to automatically computed features<br />20<br />
    • 21. Affinity Diagram Method<br />21<br />
    • 22. Endogenous vs. Exogenous Trends<br />Endogenous Trends: Twitter-centric activities that do not correspond to external events (e.g., a popular post by a celebrity)<br />Exogenous Trends: trending eventsthat originated outside of the Twitter system (e.g., an earthquake)<br />Do exogenous and endogenous trends exhibit different characteristics?<br />22<br />
    • 23. Characterization of Trends and Trending Events <br />Automatically computed features<br />Content Features<br />Interaction Features<br />Time-based Features<br />Participation Features<br />Social Network Features<br />Compared differences between categories<br />Hypotheses guided by differences in categories according to feature types<br />Performed t-tests for significance analysis<br />23<br />
    • 24. Contributions of the Study<br />Trends fall into two main categories: exogenous (i.e., trending event) and endogenous (i.e., platform-centric trend)<br />There are significant differences between exogenous and endogenous trends<br />Proportion of messages with URLs<br />Unique hashtag in top 10% of messages<br />Proportion of retweets<br />Reciprocity<br />24<br />
    • 25. Identification and Characterization of Events in Social Media<br />25<br />Characterizationof trending events <br />Identification of trending events <br />Similarity metric learning for trending events<br />Identification of content for planned events<br />Selection of event content<br />
    • 26. Identifying Trending Events<br />Event Clusters<br />Documents<br />Document Clusters<br />26<br />
    • 27. Identifying Trending Events in Real-Time<br />Order documents by post time<br />Use tf-idf vector representation of textual content<br />Stop word elimination<br />Stemming<br />idf computed over past data<br />Separate tweets by location<br />Focus on tweets from NYC<br />Different locations can be processed in parallel<br />27<br />
    • 28. Clustering Algorithm<br />Many alternatives possible! [Berkhin 2002]<br />Single-pass incremental clustering algorithm<br />Scalable, online solution<br />Using centroid representation <br />Used effectively for <br />Event identification in textual news [Allan et al. 1998]<br />News event detection on Twitter [Sankaranarayanan et al. 2009]<br />Does not require a priori knowledge of number of clusters<br />Parameters:<br />Similarity Function σ<br />Threshold μ<br />28<br />
    • 29. Overview of Cluster-based Approach<br />Group similar documents via online clustering<br />Compute statistics of cluster content <br />Top terms (e.g., [earthquake, japan])<br />Number of documents per hour<br />…<br />Use cluster-level features to identify trendingeventclusters<br />Single feature with threshold (e.g., increase in volume over time-window [Petrovićet al. 2010])<br />Trained classification model<br />29<br />
    • 30. Event Classification on Twitter<br />Cluster-level features<br />Social interaction <br />Topic coherence<br />Trending behavior<br />Platform-centric <br />Event classifier<br />Human-annotated training data<br />SVM model (selected during training phase)<br />30<br />
    • 31. Experimental Setup<br />Classification accuracy<br />Baseline: Naïve Bayes text classification (NB-Text) [Sankaranarayanan et al. 2009]<br />10-fold cross validation<br />Blind test set of randomly chosen tweets<br />Event surfacing: select top event clusters per hour<br />Baselines<br />Fastest-growing clusters per hour (Fastest) [Petrović et al. 2010]<br />Randomly selected clusters per hour (Random)<br />5 hours, top-20 clusters per hour<br />31<br />
    • 32. Identified Events<br />32<br />A sample of events identified by our classifiers on the test set<br />
    • 33. Classification Performance (F-measure)<br />RW-Event event classifier is more effective at discriminating between real-world events and rest of Twitter data<br />33<br />
    • 34. NDCG@K Evaluation<br />34<br />Performance of event classifier and baselines for event surfacing task. <br />
    • 35. Identification and Characterization of Events in Social Media<br />35<br />Characterizationof trending events <br />Identification of trending events <br />Similarity metric learning for trending events<br />Identification of content for planned events<br />Selection of event content<br />
    • 36. Social Media Document Representation<br />Title<br />Description<br />Tags<br />Date/Time<br />Location<br />All-Text<br />36<br />36<br />
    • 37. Social Media Document Similarity<br />Text: cosine similarity of tf-idf vectors (tf-idf version?; stemming?; stop word elimination?)<br />37<br />Title<br />A<br />A<br />A<br />B<br />B<br />B<br />Description<br />Time: proximity in minutes<br />Tags<br />time<br />Date/Time<br />Location: geo-coordinate proximity<br />Location<br />All-Text<br />37<br />
    • 38. Clustering Algorithm<br />Many alternatives possible! [Berkhin 2002]<br />Single-pass incremental clustering algorithm<br />Scalable, online solution<br />Using centroid representation <br />Used effectively for <br />Event identification in textual news [Allan et al. 1998]<br />News event detection on Twitter [Sankaranarayanan et al. 2009]<br />Does not require a priori knowledge of number of clusters<br />Parameters:<br />Similarity Function σ<br />Threshold μ<br />38<br />
    • 39. Cluster Representation and Parameter Tuning<br />Centroid cluster representation<br />Average tf-idf scores<br />Average time<br />Geographic mid-point<br />Parameter tuning in supervised training phase<br />Clustering quality metrics to optimize:<br />Normalized Mutual Information (NMI) [Amigó et al. 2008]<br />B-Cubed [Strehl et al. 2002]<br />39<br />
    • 40. Learning a Similarity Metric for Clustering<br />Ensemble-based similarity<br />Training a cluster ensemble<br />Computing a similarity score by:<br />Combining individual partitions<br />Combining individual similarities<br />Classification-based similarity<br />Training data sampling strategies<br />Modeling strategies<br />40<br />
    • 41. Overview of a Cluster Ensemble Algorithm<br />Ctitle<br />Ensemble clustering solution<br />Consensus Function:<br />combine ensemble <br />similarities<br />Wtitle<br />f(C,W)<br />Wtags<br />Ctags<br />Wtime<br />Ctime<br />Learned in a training step<br />41<br />
    • 42. Overview of a Cluster Ensemble Algorithm: Combining Partitions<br />Wtitle<br />Ctitle<br />f(C,W)<br />Wtags<br />Ctags<br />Wtime<br />Ctime<br />42<br />
    • 43. Overview of a Cluster Ensemble Algorithm: Combining Similarities<br />For each document di<br />and cluster cj<br />σCtitle(di,cj)>μCtitle<br />Wtitle<br />f(C,W)<br />Wtags<br />σCtags(di,cj)>μCtags<br />Wtime<br />σCtime(di,cj)>μCtime<br />43<br />
    • 44. Learning a Similarity Metric for Clustering<br />Ensemble-based similarity<br />Training a cluster ensemble<br />Computing a similarity score by:<br />Combining individual partitions<br />Combining individual similarities<br />Classification-based similarity<br />Training data sampling strategies<br />Modeling strategies<br />44<br />
    • 45. Classification-based Similarity Metrics<br />Classify pairs of documents as similar/dissimilar<br />Feature vector<br />Pairwise similarity scores <br />One feature per similarity metric (e.g., time-proximity, location-proximity, …)<br />Modeling strategies<br />Document pairs <br />Document-centroid pairs<br />45<br />
    • 46. Experiments: Alternative Similarity Metrics<br />Ensemble-based techniques<br />Combining individual partitions (ENS-PART)<br />Combining individual similarities (ENS-SIM)<br />Classification-based techniques<br />Modeling: document-document vs. document-centroidpairs<br />Logistic Regression (CLASS-LR), Support Vector Machines (CLASS-SVM)<br />Baselines<br />Title, Description, Tags, All-Text, Time-Proximity, Location-Proximity<br />46<br />
    • 47. Experimental Setup<br />Datasets:<br />Upcoming<br />>270K Flickr photos<br />Event labels from the “upcoming” event database (upcoming:event=12345)<br />Split into 3 parts for training/validation/testing<br />LastFM<br />>594K Flickr photos<br />Event labels from last.fm music catalog (lastfm:event=6789)<br />Used as an additional test set<br />47<br />
    • 48. Clustering Accuracy over Upcoming Test Set<br />All similarity learning techniques outperform the baselines<br />Classification-based techniques perform better than ensemble-based techniques<br />48<br />
    • 49. NMI: Clustering Accuracy over Both Test Sets<br /> Upcoming LastFM<br />49<br />NMI<br />Similarity learning models trained on Upcoming data show similar trends when tested on LastFM data<br />
    • 50. Identification and Characterization of Events in Social Media<br />50<br />Characterizationof trending events <br />Identification of trending events <br />Similarity metric learning for trending events<br />Identification of content for planned events<br />Selection of event content<br />
    • 51. Identifying Content for Planned Events<br />Identify planned event documents given known event information<br />User-contributed planned event records<br />LastFM Events<br />EventBrite<br />Facebook Events<br />Structured features (e.g., title, time, location)<br />Challenging identification scenario<br />Known event information is often inaccurate or incomplete<br />Social media documents are brief and noisy<br />51<br />
    • 52. Planned Event Record<br />52<br />Title<br />Description<br />Date/Time<br />Venue<br />City<br />
    • 53. Approach for Known Identification Scenario<br />Two-step query formulation strategy<br />Precision-oriented queries using known event features<br />Recall-oriented queries using retrieved content from precision-oriented queries<br />Leverage cross-site content<br />Identify event documents on each site individually<br />Use event documents on one site to retrieve additional event documents on a different site<br />53<br />
    • 54. Query Formulation Strategies<br />Precision-oriented Queries: <br />Combined event record features<br />Phrase, bag-of-words, stop word elimination<br />Examples: [“title”+”venue”], [title-no-stopwords+”city”]<br />Recall-oriented Queries<br />Frequency Analysis<br />Frequent terms in the event’s retrieved content<br />Infrequently found in Web documents<br />Term Extraction<br />54<br />
    • 55. Leveraging Cross-Site Content<br />Build precision-oriented queries using planned eventfeatures<br />Use precision-oriented queries to retrieve data from:<br />Twitter<br />Flickr<br />YouTube<br />Build recall-oriented queries using data from:<br />Each site individually<br />All sites collectively<br />55<br />[title+city]<br />[title+venue]<br />…<br />tweet1<br />tweet2<br />tweetn<br />photo1<br />photo2<br />photon<br />video1<br />video2<br />videon<br />
    • 56. Experimental Settings<br />60 planned events from EventBrite, LastFM, LinkedIn, and Facebook<br />Corresponding social media documents<br />Retrieved from Twitter, Flickr, and YouTube<br />Ranked according to similarity to event record<br />Techniques<br />Precision: only precision-oriented queries<br />MS: precision- and recall-oriented queries selected using Microsoft n-gram probability score<br />RTR: precision- and recall-oriented queries selected using ratio of document frequency around the time of the event to document frequency in larger time window<br />56<br />
    • 57. NDCG Performance on Twitter<br />57<br />NDCG scores for top-k Twitter documents retrieved by <br />Precision-oriented queries (Precision), and query strategies <br />using Twitter data (Twitter-RTR, Twitter-MS).<br />
    • 58. Cross-Site NDCG Performance<br />58<br />NDCG scores for top-k YouTube documents retrieved by <br />Precision-oriented queries (Precision), and query strategies <br />using data from Twitter (Twitter-MS) and YouTube (YouTube MS).<br />
    • 59. Identification and Characterization of Events in Social Media<br />59<br />Characterizationof trending events <br />Identification of trending events <br />Similarity metric learning for trending events<br />Identification of content for planned events<br />Selection of event content<br />
    • 60. Event Content Selection<br />60<br /> Tiger Woods to make a public apology Friday and talk about his future in golf.<br />Tiger woods y'all,tiger woods y'all,ah tiger woods y'all<br />Tiger Woods Apology<br />Tiger Woods Hugs: http://tinyurl.com/yhf4uzw<br />Tiger Woods Returns To Golf - Public Apology http://bit.ly/9Ui5jx<br />Wedge wars upstage Watson v Woods: BBC Sport (blog)<br />
    • 61. Event Content Selection<br />Challenges:<br />Document clusters contain noise<br />Relevant documents might have poor quality text<br />Relevant, high quality documents might not be interesting<br />For each document and a given event evaluate<br />Quality<br />Relevance<br />Usefulness<br />61<br />
    • 62. Centrality Based Document Selection<br />Centroid<br />Cosine similarity of each document to cluster centroid<br />Degree<br />Documents are nodes<br />Documents are connected if their similarity is above a threshold<br />Compute degree centrality of each node<br />LexRank[Erkan and Radev 2004]<br />Same graph structure as Degree method <br />Central documents are similar to other central documents<br />62<br />
    • 63. Experimental Methodology: Content Selection<br />50 event clusters<br />Randomly selected<br />5 top tweets per event for each: Centroid, Degree, LexRank<br />Labeled on a 1-4 scale<br />Quality: excellent (4) poor (1)<br />Relevance: clearly relevant (4)  not relevant (1)<br />Usefulness: clearly useful (4)  not useful (1)<br />63<br />
    • 64. Content Selection Results<br />Average scores over all events (out of 4)<br />High quality and relevance (>3) for both Degree and Centroid<br />Centroid only method with high usefulness <br />64<br />
    • 65. Conclusions<br />Techniques for identifying, characterizing, and selecting social media content for events<br />There are significant differences between types of trends in social media, specifically trending events and non-event trends<br />Trending events and their associated social media documents can be effectively identified using online clustering with:<br />A classification step to separate event and non-event content<br />Social media document similarity metrics for documents with rich context features<br />A two-step query formulation technique is useful for identifying planned events across different social media sites <br />Centrality-based techniques can be used to select high quality, relevant, and useful social media event content<br />65<br />
    • 66. Future Work<br />Clustering framework optimization<br />Blocking techniques<br />Topic models<br />Identify unknown events with learned similarity metrics across sites<br />Improve breadth of event content<br />Rank events for search and presentation<br />Extension of content selection techniques<br />Learned ranking models<br />66<br />
    • 67. Publications<br />Hila Becker, Dan Iter, MorNaaman, Luis Gravano, “Identifying Content for Planned Events Across Social Media Sites,” under submission.<br />Hila Becker, Mor Naaman, Luis Gravano, “Beyond Trending Topics: Real-World Event Identification on Twitter,” in Proceedings of the Fifth International AAAI Conference on Weblogs and Social Media (ICWSM’11), short paper.<br />Hila Becker, Mor Naaman, Luis Gravano, “Selecting Quality Twitter Content for Events,” in Proceedings of the Fifth International AAAI Conference on Weblogs and Social Media (ICWSM’11), short paper.<br />Hila Becker, Feiyang Chen, Dan Iter, Mor Naaman, Luis Gravano, “Automatic Identification and Presentation of Twitter Content for Planned Events,” in Proceedings of the Fifth International AAAI Conference on Weblogs and Social Media (ICWSM’11), demo paper.<br />Mor Naaman, Hila Becker, Luis Gravano, “Hip and Trendy: Characterizing Emerging Trends on Twitter,” in Journal of the American Society for Information Science and Technology. <br />Hila Becker, MorNaaman, Luis Gravano, “Learning Similarity Metrics for Event Identification in Social Media,” in Proceedings of the Third ACM International Conference on Web Search and Data Mining (WSDM '10), 291-300.<br />Hila Becker, Bai Xiao, MorNaaman and Luis Gravano, “Exploiting Social Links for Event Identification in Social Media,” in Proceedings of the 3rd Annual Workshop on Search in Social Media (SSM '10), poster paper.<br />Hila Becker, MorNaaman, Luis Gravano, “Event Identification in Social Media,” in Proceedings of the ACM SIGMOD Workshop on the Web and Databases (WebDB '09), 2009.<br />67<br />
    • 68. Thank You!<br />68<br />
    • 69. 69<br />

    ×