Keynote Title: Retrieving Knowledge from the Web
We all turn towards Wikipedia with questions we want to know more about, but eventually find ourselves on the limit of its coverage. Instead of providing "ten blue links" as common in Web search, why not answer any web query with something that looks and feels like Wikipedia? This talk is about algorithms that automatically retrieve, extract, and compile a knowledge resource for a given web query. The trick is to model the duality between structured knowledge and unstructured text. This leads to supervised retrieval models can jointly identify relevant Web documents, Wikipedia entities, and extract support passages to populate knowledge articles.
Laura Dietz is an Assistant Professor at University of New Hampshire where she teaches Information Retrieval and text-centric Machine Learning. Before that she was working as a research scientist in the at the Data and Web science group at Mannheim University, with Bruce Croft and Andrew McCallum at the Center for Intelligent Information Retrieval
(CIIR) at the University of Massachusetts. She obtained her doctoral degree with a thesis on topic models for networked data from Max Planck Institute for Informatics, being supervised by Tobias Scheffer and Gerhard Weikum.
Keynote Title: Efficient query results on the content of millions of videos
Even though the accuracy of content based video search systems (CBVS) has drastically improved in recent years, high accuracy video retrieval systems tend to be too inefficient for interactive search. Therefore, to achieve real-time CBVS over millions of videos, we perform a comprehensive study on the different components in a CBVS system to understand the tradeoffs between accuracy and speed of each component. Directions investigated include exploring different low-level and semantics based features, testing different compression factors and approximations during video search, and understanding the time vs. accuracy trade-of. Semantic search in video is a novel and challenging problem in information and multimedia retrieval. Existing solutions are mainly limited to text matching, in which the query words are matched against the textual metadata generated by users. This talk
will contrast approaches for content search both with example videos and without, using only text queries. The system relies on substantial video content analysis and allows for both low-level and semantic search over a large collection of videos. We share our observations and lessons in building such a system, which may be instrumental in guiding the design of future systems for search in video. Extensive experiments on very large archives consisting of more than 2,000 hours of short videos showed that through a combination of effective features, highly compressed representations, and reranking, our proposed system can achieve an 10,000-fold speedup while retaining 80% accuracy of a state-of-the-art CBVS system. Over 1 million videos, our system can complete exemplar-based search in one second with a single core. A text-only search solution with slightly lower accuracy easily scaled to 100 million videos.
Alex Hauptmann in currently Principal Systems Scientist in CMU School of Computer Science's Language Technologies Institute. He received his B.A. and M.A. in Psychology from Johns Hopkins University, a Diplom (M.S.) in Computer Science at the Technische Universität Berlin, and received his Ph.D. in Computer Science from CMU in 1991. His research interests have led him to pursue and combine several different areas: man-machine communication, natural language processing, speech understanding and synthesis, machine learning. He worked on speech and machine translation at CMU from 1984-94, when he became leader of the Informedia project. He has worked for over 20 years in video analysis, video retrieval, video summarization and interfaces for video search systems. He has authored and co-authored over 250 refereed papers which have been cited more than 13000 times. He was a PI on projects for the IARPA (ARDA/DTO) VACE, AQUAINT programs and most recently the ALADDIN program, which extended the Informedia digital video library system into more detailed visual analysis (VACE, ALADDIN), as well as investigating search approaches as a form of video question answering (AQUAINT). In addition, he has been a PI on a number of DARPA, NSF and NIH projects related to video understanding, semantic indexing of video and data mining of video content. His research success on video analysis and retrieval is documented by outstanding performance in competitions such as the annual NIST TRECVID video retrieval evaluation and the CVPR Thumos activity recognition challenge.
Keynote: People are creatures of habit. We take the same route into work every day, email the same people, and visit the same webpages over and over again. So although search engines are targeted towards helping people find new information, it is not surprising that people also regularly use them to re-find content that they have seen before. This creates a rich opportunity for us to take advantage of re-finding behavior as a means to improve the search experience. This talk will provide an overview of recent research on re-searching, characterizing people's search habits and suggesting ways that their habits can be used to help them find what they are looking for. It will also highlight potential pitfalls that can be avoided by acknowledging that people have prior experiences and expectations that they bring with them every time they search.
Biography: Jaime Teevan is a Principal Researcher at Microsoft Research and an Affiliate Associate Professor at the University of Washington. At Microsoft Research she leads the microproductivity team and shipped the first personalized search algorithm used by Bing.
Dr. Teevan has published hundreds of award-winning research papers, technical articles, books, and patents, and given keynotes and lectures around the world. Her groundbreaking research at the intersection of information retrieval and human computer interaction has earned her the Technology Review TR35 Young Innovator, Borg Early Career, and Karen Spärck Jones awards. She received a Ph.D. and S.M. from MIT and a B.S. with honors from Yale University. http://teevan.org