View Document Preview and Link Document Date: 2012-10-30 19:13:40 Open Document File Size: 1,75 MB Share Result on Facebook
City Pittsburgh / / Company X Lc / 3M / Intel / Microsoft / / Country United States / / / Facility Jagged Column Store / Column Store / Yucheng Low Carnegie Mellon University / CS store / In-Memory Matrix Representations Jagged Column Store / Jagged Row And Column Store / only be implemented using the R&CS store / / IndustryTerm user search / final pre-processing stage / actual inner product / query processing / inner product / online advertising / mWand algorithm / sketch algorithm / contiguous / inner products / binary search trees / multi-step k-nearest neighbor search / pairs similarity search / baseline naive algorithm / word-word inner product / sparse inner products / matrix vector product / similarity search / search order / hybrid algorithms / on-line advertising / large search engines / search heuristic / binary search / on-line incremental construction / vector inner products / naive algorithm / mWand iterator algorithm / slowest algorithms / native algorithm / streaming/online insertion / inner product search / search orientation / Near-optimal hashing algorithms / approximate solutions / / MarketIndex set 5000 / / Organization idf / U.S. Securities and Exchange Commission / K. W. Church / Carnegie Mellon University / Association for Computational Linguistics / / Person H. Daumé / III / J. Jagarlamudi / M. Fontoura / V / A. André F. T. Martins / S. Venkatasubramanian / A. Goyal / Alice X. Zheng / / / Position Linguist / / ProvinceOrState Pennsylvania / / PublishedMedium Computational Linguistics / Machine Learning / / TVStation Wand / / Technology mWand iterator algorithm / Naive R&CS algorithm / RAM / implemented algorithms / 2.4 HComp Algorithm / slowest algorithms / Machine Learning / baseline naive algorithm / HComp CS algorithm / Wand algorithm / HComp algorithm / mWand algorithm / naive algorithm / TAAT algorithms / same algorithm / sketch algorithm / native algorithm / / SocialTag 
Fast Top-K Similarity Queries Via Matrix Compression ∗ Yucheng Low Carnegie Mellon University