WePS: searching information about entities in the Web

  • Increase font size
  • Default font size
  • Decrease font size

WePS-3 Workshop Program

9:30-11:00WePS overviewTask-1: Clustering
session chair: Satoshi Sekine
  • WePS overview (Julio Gonzalo)
  • Task 1 overview (Javier Artiles)
Presentations by participants in Task-1 (20 min. each)
  • Web Person Name Disambiguation by Relevance Weighting of Extended Feature Sets. Chong Long and Lei Shi.
  • Using Web Graph Structure for Person Name Disambiguation Elena Smirnova, Konstantin Avrachenkov and Brigitte Trousse,
  • Cross-document coreference for WePS. Iustin Dornescu, Constantin Orasan and Tatiana Lesnikova.
coffee break
11:30-12:00Task-1: Clustering (cont.)
session chair: Satoshi SekinePresentations by participants in Task-1 (15 min. each)
  • DAEDALUS at WePS-3 2010: k-Medoids Clustering using a Cost Function Minimization. Sara Lana-Serrano, Julio Villena-Román and José-Carlos González-Cristóbal.
  • WePS3 person attribute extraction from the textual parts of Web pages. István Nagy and Richárd Farkas
    Invited talk
    session chair: Julio Gonzalo
  • Maarten de Rijke, Sentimental About People
    The talk is a mixture of descriptive and algorithmic aspects of searching for people. I will start by surveying search scenarios in which people are the target. In several use cases, something more is desired: not just people but also expressions of opinions about people. The performance of current sentiment retrieval methods applied to people search will be described and I will detail ongoing work in which we use syntactic information in a bootstrapping fashion to improve the effectiveness of finding sentiments about people in social media.The talk is based on ongoing joint work with Krisztian Balog, Valentin Jijkoun and Wouter Weerkamp.
lunch break
14:30-16:00Task-2: Online Reputation Management
session chair: Javier Artiles
  • Task-2 overview (Enrique Amigó)
  • Mechanical Turk methodology (Andrew Borthwick)
Presentations by participants in Task-2 (15 min. each)
  • It was easy, when apples and blackberries were only fruits. Surender Reddy Yerva, Zoltá, Miklós and Karl Aberer.
  • ITC-UT:Tweet Categorization by Query Categorization of On-line Reputation Management. Minoru Yoshida, Shin Matsushima, Shingo Ono, Issei Sato, and Hiroshi Nakagawa.
  • The University of Amsterdam in WePS3. Manos Tsagkias and Krisztian Balog.
  • Bootstrapping Websites for Classification of Organization Names on Twitter. Paul Kalmar.
coffee break
16:30-17:30Invited talk
session chair: Andrew Borthwick
  • Invited talk by Heng Ji, Overview of the TAC2010 Knowledge Base Population Track.
    In this talk I will give an overview of the Knowledge Base Population (KBP) track at TAC 2010. The main goal of KBP is to promote research in discovering facts about entities and expanding a knowledge source. A large source collection of newswire and web documents is provided for systems to discover information. Attributes (a.k.a., "slots") derived from Wikipedia infoboxes are used to create the reference knowledge base (KB). KBP2010 includes the following four tasks: (1). Regular Entity Linking, where names must be aligned to entities in the KB; (2). Optional Entity linking, without using Wikipedia texts; (3). Regular Slot Filling, which requires a system to automatically discover answers from the source document collection and return answers for a query entity with specified slots, and use them to expand KB; (4). Surprise Slot Filling, which requires a system to return answers regarding new slot types. In addition, a community effort was organized to create slot filling training corpora. KBP2010 has attracted many participants from Information Extraction (IE) and Question Answering (QA) communities (54 teams registered, 24 teams submitted results). In this talk I present the annotation and implementation challenges associated with KBP2010. Then I will briefly present a state-of-the-art slot filling system that combines IE and QA approach, and exploits statistical answer Re-ranking and Markov Logic Networks for Cross-slot reasoning. Furthermore, I will present an approach of combining answers across top systems and human annotators, and show that filtering errors from system combination can achieve higher recall and save annotation cost than traditional human annotations from scratch. At the end I will also describe some possible new tasks in KBP 2011 including cross-lingual KBP and temporal KBP.
17:30-18:00Discussion on the future of WePS
moderator: Andrew Borthwick