Low-rank approximations of second-order document representations

Show full item record



Permalink

http://hdl.handle.net/10138/309458

Citation

Lagus , J , Sinkkonen , J & Klami , A 2019 , Low-rank approximations of second-order document representations . in M Bansal & A Villavicencio (eds) , Proceedings of the 23rd Conference on Computational Natural Language Learning (CoNLL) . ACL , Stroudsburg, PA , pp. 634-644 , Conference on Computational Natural Language Learning , Hong Kong , Hong Kong , 03/11/2019 . https://doi.org/10.18653/v1/K19-1059

Title: Low-rank approximations of second-order document representations
Author: Lagus, Jarkko; Sinkkonen, Janne; Klami, Arto
Other contributor: University of Helsinki, Reaktor Oy
University of Helsinki, Department of Computer Science
Bansal, Mohit
Villavicencio, Aline

Publisher: ACL
Date: 2019-11
Language: eng
Number of pages: 11
Belongs to series: Proceedings of the 23rd Conference on Computational Natural Language Learning (CoNLL)
ISBN: 978-1-950737-72-7
DOI: https://doi.org/10.18653/v1/K19-1059
URI: http://hdl.handle.net/10138/309458
Abstract: Document embeddings, created with methods ranging from simple heuristics to statistical and deep models, are widely applicable. Bag-of-vectors models for documents include the mean and quadratic approaches (Torki, 2018). We present evidence that quadratic statistics alone, without the mean information, can offer superior accuracy, fast document comparison, and compact document representations. In matching news articles to their comment threads, low-rank representations of only 3-4 times the size of the mean vector give most accurate matching, and in standard sentence comparison tasks, results are state of the art despite faster computation. Similarity measures are discussed, and the Frobenius product implicit in the proposed method is contrasted to Wasserstein or Bures metric from the transportation theory. We also shortly demonstrate matching of unordered word lists to documents, to measure topicality or sentiment of documents.
Subject: 113 Computer and information sciences
Rights:


Files in this item

Total number of downloads: Loading...

Files Size Format View
K19_1059.pdf 2.350Mb PDF View/Open

This item appears in the following Collection(s)

Show full item record