• Publications
  • Influence
Knowledge-Based Weak Supervision for Information Extraction of Overlapping Relations
TLDR
A novel approach for multi-instance learning with overlapping relations that combines a sentence-level extraction model with a simple, corpus-level component for aggregating the individual facts is presented. Expand
Design and Analysis of Contracts for Software Outsourcing
TLDR
A contract-theoretic model is presented that finds that despite their relative inefficiency, fixed-price contracts are often appropriate for simple software projects that require short development time and time-and-materials contracts work well for more complex projects when the auditing process is efficient and effective. Expand
Learning 5000 Relational Extractors
TLDR
LUCHS is presented, a self-supervised, relation-specific IE system which learns 5025 relations --- more than an order of magnitude greater than any previous approach --- with an average F1 score of 61%. Expand
Harvesting Parallel News Streams to Generate Paraphrases of Event Relations
TLDR
Three Temporal Correspondence Heuristics, that characterize regularities in parallel news streams, are introduced, and it is shown how they may be used to generate high precision paraphrases for event relations. Expand
Exploiting Parallel News Streams for Unsupervised Event Extraction
TLDR
NewsSpike-RE is introduced, a novel, unsupervised algorithm that discovers event relations and then learns to extract them, more than doubling the area under a precision-recall curve compared to Universal Schemas. Expand
Machine Reading at the University of Washington
TLDR
A unifying approach for machine reading is proposed by bootstrapping from the easiest extractable knowledge and conquering the long tail via a self-supervised learning process that is made scalable by leveraging hierarchical structures and coarse-to-fine inference. Expand
Adaptive Parser-Centric Text Normalization
TLDR
This paper takes a parser-centric view of normalization that aims to convert raw informal text into grammatically correct text, and demonstrates that this approach outperforms not only the state-of-the-art wordto-word normalization techniques, but also manual word-to- word annotations. Expand
A Novel Web Page Categorization Algorithm Based on Block Propagation Using Query-Log Information
TLDR
A Block Propagation Categorization (BPC) algorithm which deep mines web structure and views blocks as basic semantic units and propagates only suitable information (block) among web pages to emphasize their topics. Expand
Ontological Smoothing for Relation Extraction with Minimal Supervision
TLDR
Experiments on 65 relations across three target domains show that ontological smoothing can dramatically improve precision and recall, even rivaling fully supervised performance in many cases. Expand
Web-scale classification with naive bayes
TLDR
Modifications to the traditional Naive Bayes Classifier are proposed that can alleviate the contradiction pair problem and discriminative evidence cancelation problem and significantly improve the performance on real web-scale taxonomies. Expand
...
1
2
...