Distant supervision relationship extractor

Abstract

Automatic relationship extraction is provided. A machine learning approach using statistical entity-type prediction and relationship predication models built from large unlabeled datasets is interactively combined with minimal human intervention and a light pattern-based approach to extract relationships from unstructured, semi-structured, and structured documents. Training data is collected from a collection of unlabeled documents by matching ground truths for a known entity from existing fact databases with text in the documents describing the known entity and corresponding models are built for one or more relationship types. For a modeled relationship-type, text chunks of interest are found in a document. A machine learning classifier predicts the probability that one of the text chunks is the entity being sought. The combined machine learning and light pattern-based approach provides both improved recall and high precision through filtering and allows constraining and normalization of the extracted relationships.

Description

Background technique [0001] Populating a database of facts describing the relationships between entities and their attributes typically requires the aggregation of a lot of information with a high level of precision. Populating large fact databases manually is time-consuming, expensive, and often impractical. Automatically populating fact databases can also be a challenge due to the difficulty in extracting data with the necessary precision from varying sources of structured, semi-structured and unstructured information using inconsistent language, units and formats without human supervision. time consuming. Conventional automatic fact extraction techniques include pattern matching and natural language processing. [0002] Pattern matching typically uses hand-crafted and hard-coded regular expressions and / or specific rules that rely on information being expressed in the same order using the same words. Without a comprehensive schema set, many expressions of relationships ca...

AI Technical Summary

Problems solved by technology

Adding more patterns can reduce the number of missed expressions and may lead to collection of irrelevant data

Ultimately, while careful pattern matching may improve, creating patterns is time-consuming, expensive, and not scalable

[0003] Natural language processing using statistical models is not bound by a particular paradigm, but building good models requires lots of properly annotated training data

Manually annotating large datasets to build high-precision models is time-consuming and expensive

Images