Hong-Woo Chun

Extraction of gene-disease relations from Medline using domain dictionaries and machine learning.

We describe a system that extracts disease-gene relations from Medline. We constructed a dictionary for disease and gene names from six public databases and extracted relation candidates by dictionary matching. Since dictionary matching produces a large number of false positives, we developed a method of machine learning-based named entity recognition (NER) to filter out false recognitions of disease/gene names. We found that the performance of relation extraction is heavily dependent upon the performance of NER filtering and that the filtering improves the precision of relation extraction by 26.7% at the cost of a small reduction in recall.

A Markov Logic Approach to Bio-Molecular Event Extraction

In this paper we describe our entry to the BioNLP 2009 Shared Task regarding biomolecular event extraction. Our work can be described by three design decisions: (1) instead of building a pipeline using local classifier technology, we design and learn a joint probabilistic model over events in a sentence; (2) instead of developing specific inference and learning algorithms for our joint model, we apply Markov Logic, a general purpose Statistical Relation Learning language, for this task; (3) we represent events as relational structures over the tokens of a sentence, as opposed to structures that explicitly mention abstract event entities. Our results are competitive: we achieve the 4th best scores for task 1 (in close range to the 3rd place) and the best results for task 2 with a 13 percent point margin.

Automatic recognition of topic-classified relations between prostate cancer and genes using MEDLINE abstracts

BackgroundAutomatic recognition of relations between a specific disease term and its relevant genes or protein terms is an important practice of bioinformatics. Considering the utility of the results of this approach, we identified prostate cancer and gene terms with the ID tags of public biomedical databases. Moreover, considering that genetics experts will use our results, we classified them based on six topics that can be used to analyze the type of prostate cancers, genes, and their relations.MethodsWe developed a maximum entropy-based named entity recognizer and a relation recognizer and applied them to a corpus-based approach. We collected prostate cancer-related abstracts from MEDLINE, and constructed an annotated corpus of gene and prostate cancer relations based on six topics by biologists. We used it to train the maximum entropy-based named entity recognizer and relation recognizer.ResultsTopic-classified relation recognition achieved 92.1% precision for the relation (an increase of 11.0% from that obtained in a baseline experiment). For all topics, the precision was between 67.6 and 88.1%.ConclusionA series of experimental results revealed two important findings: a carefully designed relation recognition system using named entity recognition can improve the performance of relation recognition, and topic-classified relation recognition can be effectively addressed through a corpus-based approach using manual annotation and machine learning techniques.

U-Compare bio-event meta-service: compatible BioNLP event extraction services

BACKGROUND Bio-molecular event extraction from literature is recognized as an important task of bio text mining and, as such, many relevant systems have been developed and made available during the last decade. While such systems provide useful services individually, there is a need for a meta-service to enable comparison and ensemble of such services, offering optimal solutions for various purposes. RESULTS We have integrated nine event extraction systems in the U-Compare framework, making them intercompatible and interoperable with other U-Compare components. The U-Compare event meta-service provides various meta-level features for comparison and ensemble of multiple event extraction systems. Experimental results show that the performance improvements achieved by the ensemble are significant. CONCLUSIONS While individual event extraction systems themselves provide useful features for bio text mining, the U-Compare meta-service is expected to improve the accessibility to the individual systems, and to enable meta-level uses over multiple event extraction systems such as comparison and ensemble.

Procedural knowledge extraction on MEDLINE abstracts

Text mining is a popular methodology for building Technology Intelligence which helps companies or organizations to make better decisions by providing knowledge about the state-of-the-art technologies obtained from the Internet or inside companies. As a matter of fact, the objects or events (socalled declarative knowledge) are the target knowledge that text miners want to catch in general. However, we propose how to extract procedural knowledge rather than declarative knowledge utilizing machine learning method with deep language processing features, as well as how to model it. We show the representation of procedural knowledge in MEDLINE abstracts and provide experiments that are quite promising in that it shows 82% and 63% performances of purpose/solutions (two components of procedural knowledge model) extraction and unit process (basic unit of purpose/solutions) identification respectively, even though we applied strict guidelines in evaluating the performance.

Multi-words Terminology Recognition Using Web Search

Terminology recognition system which is a fundamental research for Technology Opportunity Discovery (TOD) has been intensively studied in limited range of domains, especially in bio-medical domain. We propose a domain independent terminology recognition system based on machine learning method using dictionary, syntactic features, and Web search results, since the previous works revealed limitation on applying their approaches to general domain because their resources were domain specific. We achieved F-score 80.4 and 6.4% improvement after comparing the proposed approach with the related approach, C-value, which has been widely used and is based on local domain frequencies. In the second experiment with various combinations of unithood features, the method combined with NGD(Normalized Google Distance) showed the best performance of 81.5 on F-score. We applied two machine learning methods such as Logistic regression and SVMs, and got the best score at SVMs method.

BIO‐MOLECULAR EVENT EXTRACTION WITH MARKOV LOGIC

This article presents a novel approach to event extraction from biological text using Markov Logic. It can be described by three design decisions: (1) instead of building a pipeline using local classifiers, we design and learn a joint probabilistic model over events in a sentence; (2) instead of developing specific inference and learning algorithms for our joint model, we apply Markov Logic, a general purpose Statistical Relation Learning language, for this task; (3) we represent events as relations over the token indices of a sentence, as opposed to structures that relate event entities to gene or protein mentions. In this article, we extend our original work by providing an error analysis for binding events. Moreover, we investigate the impact of different loss functions to precision, recall and F‐measure. Finally, we show how to extract events of different types that share the same event clue. This extension allowed us to improve our performance our performance even further, leading to the third best scores for task 1 (in close range to the second place) and the best results for task 2 with a 14% point margin.

Relation Extraction Based on Composite Kernel Combining Pattern Similarity of Predicate-Argument Structure

Lots of valuable textual information is used to extract relations between named entities from literature. Composite kernel approach is proposed in this paper. The composite kernel approach calculates similarities based on the following information:(1) Phrase structure in convolution parse tree kernel that has shown encouraging results. (2) Predicate-argument structure patterns. In other words, the approach deals with syntactic structure as well as semantic structure using a reciprocal method. The proposed approach was evaluated using various types of test collections and it showed the better performance compared with those of previous approach using only information from syntactic structures. In addition, it showed the better performance than those of the state of the art approach.

Smart searching system for virtual science brain

To decide research topics or analyze technical trends, researchers should collect and analyze information from hundreds of thousands of articles, patents, and technical reports. To facilitate the process, information extraction techniques from literature are very helpful. In addition, effective searching methods of the extracted information are necessary as well. While information extraction research has been a popular issue, research about searching and browsing methods for the extracted information has not been an attractive issue relatively. This paper presents a smart searching system that provides various analysis tools, and we expect that researchers can discover and develop new research outcomes through the proposed searching system.

Terminology Recognition System based on Machine Learning for Scientific Document Analysis

Terminology recognition system which is a preceding research for text mining, information extraction, information retrieval, semantic web, and question-answering has been intensively studied in limited range of domains, especially in bio-medical domain. We propose a domain independent terminology recognition system based on machine learning method using dictionary, syntactic features, and Web search results, since the previous works revealed limitation on applying their approaches to general domain because their resources were domain specific. We achieved F-score 80.8 and 6.5% improvement after comparing the proposed approach with the related approach, C-value, which has been widely used and is based on local domain frequencies. In the second experiment with various combinations of unithood features, the method combined with NGD(Normalized Google Distance) showed the best performance of 81.8 on F-score. We applied three machine learning methods such as Logistic regression, C4.5, and SVMs, and got the best score from the decision tree method, C4.5.