Relation extraction and scoring in DeepQA

@article{Wang2012RelationEA,
  title={Relation extraction and scoring in DeepQA},
  author={Chang Wang and Aditya Kalyanpur and James Fan and Branimir Boguraev and David Gondek},
  journal={IBM J. Res. Dev.},
  year={2012},
  volume={56},
  pages={9}
}
Detecting semantic relations in text is an active problem area in natural-language processing and information retrieval. [] Key Method This paper presents two approaches to broad-domain relation extraction and scoring in the DeepQA question-answering framework, i.e., one based on manual pattern specification and the other relying on statistical methods for pattern elicitation, which uses a novel transfer learning technique, i.e., relation topics.

Figures and Tables from this paper

Web relation extraction with distant supervision
TLDR
This thesis explores what can cause NERC methods to fail in diverse genres and quantifies different reasons for NERC failure, and proposes solutions for issues arising for information extraction for not traditionally studied domains.
Automatic knowledge extraction from documents
TLDR
This paper describes in detail what kind of shallow knowledge is extracted, how it is automatically done from a large corpus, and how additional semantics are inferred from aggregate statistics of the automatically extracted shallow knowledge.
An Effective Corpus-Based Question Answering Pipeline for Italian
TLDR
A Question Answering pipeline for Italian and based on a corpus of documents pertaining a closed domain that exhibits functionalities for analyzing natural language questions in Italian by using lexical features and scoring and selecting candidate answers with respect to their type in order to determine the best one.
Semantic Parsing for Technical Support Questions
TLDR
An approach for semantic parsing of technical questions that uses grammatical structure to extract these attributes as a baseline, and a CRF based model that can improve performance considerably in the presence of annotated data for training are presented.
Towards a Framework for Closed-Domain Question Answering in Italian
TLDR
A preliminary Question Answering framework for closed-domains, like Cultural Heritage, that exploits a variety of NLP methods for the Italian language to help the understanding of user's questions and the extraction of precise answers from textual passages contained into documents.
Relation Extraction : A Survey
TLDR
This survey surveys several important supervised, semi-supervised and unsupervised Relation Extraction techniques and describes some of the recent trends in the RE techniques and possible future research directions.
Effective Chinese Relation Extraction by Sentence Rolling and Candidate Ranking
TLDR
This paper proposes a method called sentence rolling to generate several enhanced inputs from the original input to help generate the correct relation candidates and shows that the method can improve the performance of relation extraction significantly compared with the existing ones and cost a reasonable time.
Encyclopaedic question answering
TLDR
A novel semantic architecture is proposed which represents a paradigm shift in open-domain QA system design, using semantic concepts and knowledge representation instead of words and information retrieval.
Structured data and inference in DeepQA
TLDR
A survey of DeepQA components that use structured data is presented and details of appropriate components are presented and their end-to-end impact on the IBM Watsoni system is demonstrated.
Entity Pairs , Types and Sentences X-Entity Y-Entity
  • 2014
...
1
2
3
4
5
...

References

SHOWING 1-10 OF 68 REFERENCES
Automatic knowledge extraction from documents
TLDR
This paper describes in detail what kind of shallow knowledge is extracted, how it is automatically done from a large corpus, and how additional semantics are inferred from aggregate statistics of the automatically extracted shallow knowledge.
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%.
Relation Extraction with Relation Topics
TLDR
The experimental results on Wikipedia and ACE data have confirmed that background-knowledge-based topics generated from the Wikipedia relation repository can significantly improve the performance over the state-of-the-art relation detection approaches.
Structured data and inference in DeepQA
TLDR
A survey of DeepQA components that use structured data is presented and details of appropriate components are presented and their end-to-end impact on the IBM Watsoni system is demonstrated.
Question analysis: How Watson reads a clue
TLDR
How elements of the question that play particular roles that may require special handling, for example, nested subquestions that must be separately answered are detected and the impact of accurate detection on the end-to-end question-answering system accuracy is evaluated.
Finding needles in the haystack: Search and candidate generation
TLDR
A suite of candidate generation strategies that use various kinds of metadata, such as document titles or anchor texts in hyperlinked documents are discussed, demonstrating that a combination of these strategies brings the correct answer into the candidate answer pool for 87.17% of all the questions in a blind test set, facilitating high end-to-end question-answering performance.
Textual evidence gathering and analysis
TLDR
This work retrieves passages using a novel technique that it calls Supporting Evidence Retrieval, in which separate search queries are performed for each candidate answer, in parallel, and include the candidate answer as part of the query.
Extracting Relations with Integrated Information Using Kernel Methods
TLDR
This paper presents an evaluation of these methods on the 2004 ACE relation detection task, using Support Vector Machines, and shows that each level of syntactic processing contributes useful information for this task.
Convolution Kernels on Constituent, Dependency and Sequential Structures for Relation Extraction
TLDR
This paper explores the use of innovative kernels based on syntactic and semantic structures for a target relation extraction task and illustrates that the combination of the above kernels achieves high effectiveness and significantly improves the current state-of-the-art.
A framework for merging and ranking of answers in DeepQA
TLDR
This work describes the framework, explains the challenges, and evaluates the gain over a baseline machine learning approach, showing how this design can be used to implement solutions to particular challenges that arise in applying machine learning for evidence-based hypothesis evaluation.
...
1
2
3
4
5
...