Abstract
Querying both structured and unstructured data via a single common query interface such as SQL or natural language has been a long standing research goal. Moreover, as methods for extracting information from unstructured data become ever more powerful, the desire to integrate the output of such extraction processes with “clean”, structured data grows. We are convinced that for successful integration into databases, such extracted information in the form of “triples” needs to be both (1) of high quality and (2) have the necessary generality to link up with varying forms of structured data. It is the combination of both these aspects, which heretofore have been usually treated in isolation, where our approach breaks new ground.The cornerstone of our work is a novel, generic method for extracting open information triples from unstructured text, using a combination of linguistics and learning-based extraction methods, thus uniquely balancing both precision and recall. Our system called LILLIE (LInked Linguistics and Learning-Based Information Extractor) uses dependency tree modification rules to refine triples from a high-recall learning-based engine, and combines them with syntactic triples from a high-precision engine to increase effectiveness. In addition, our system features several augmentations, which modify the generality and the degree of granularity of the output triples. Even though our focus is on addressing both quality and generality simultaneously, our new method substantially outperforms current state-of-the-art systems on the two widely-used CaRB and Re-OIE16 benchmark sets for information extraction.We have made our code publicly available11https://github.com/OIELILLIE/LILLIE. to facilitate further research.
Highlights
It is commonly known that some 80% of enterprise data is unstructured while only some 20% is structured [1,2]
The paper is organized as follows: in Section 2 we review the related work on information extraction and entity linking for knowledge base construction; in Section 3, we give an overview of the LILLIE architecture; in Sections 4 and 5, we describe the algorithms and functions of the rule-based extractor and the learning-based extractor, respectively; in Sections 6 and 7, we show how to combine both engines, and customize their output; in Section 8, we describe how to apply our triple extractor to the task of entity linking and database insertion; in Section 9, we give a detailed analysis and evaluation of all the components of our system, and compare these to the current state-of-the-art
Entity Linking (EL) systems are capable of resolving the lexical ambiguity of entity mentions and can be extremely useful in a plethora of natural understanding (NLU) applications, by enriching the information extracted via Open Information Extraction (OIE) systems
Summary
It is commonly known that some 80% of enterprise data is unstructured while only some 20% is structured [1,2]. In order to query both structured and unstructured data via a single common query interface such as SQL or natural language [3,4], there have been several research efforts over the last years. One such approach, which we follow in our work, is to first use information extraction techniques to retrieve relevant entities (subjects and objects) and relationships. The subject ‘‘THY1’’ and the object ‘‘human gallbladder carcinoma’’ are linked to the relational database Building such an end-to-end pipeline to enable the vision of querying structured and unstructured data via a common interface has been a long standing research effort [6,7].
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
