Transcript Methodology (Cont.)

國立雲林科技大學
N.Y.U.S.T.
I. M.
National Yunlin University of Science and Technology
Psychiatric document retrieval using a
discourse-aware model
Presenter : Wu, Jia-Hao
Authors : Liang-Chih Yu, Chung-Hsien Wu , Fong-Lin Jang
Artificial Intelligence (2009)
Intelligent Database Systems Lab
Outline

Motivation

Objective

Methodology

A Framework of psychiatric document retrieval

Discourse-aware retrieval model

Experiments

Conclusion

Personal Comments
N.Y.U.S.T.
I. M.
2
Intelligent Database Systems Lab
Motivation
N.Y.U.S.T.
I. M.

Individuals in their daily life may suffer from negative or
stressful life events.

Some website provide suggestions for individuals.

Browsing and searching all consultation documents to identify the
relevant documents is time consuming and tends to become
overwhelming.
Intelligent Database Systems Lab
Objective

N.Y.U.S.T.
I. M.
The paper proposes the use of high-level discourse-aware
model.

The model can extract from queries and documents to improve the
precision of retrieval results about the psychiatric document retrieval.

Some Retrieval models , such as vector space model and Okapi model ,
but there only consider the word-level information in queries and
documents.
Consultation
Documents
Query
Recommendation
Intelligent Database Systems Lab
Methodology
N.Y.U.S.T.
I. M.
Temporal-effect
Cause-effect
Discourse = Events + Symptoms + Relations
Intelligent Database Systems Lab
Methodology (Cont.)

Negative life event identification


N.Y.U.S.T.
I. M.
Find the patterns from the sentences.
Pattern induction

Use the seed patterns from psychiatry web corpora using an
evolutionary inference algorithm.
<Husband, argue> →<Husband, fight> ,<husband, yell>,<wife, argue> , <husband, fight, money>
 SVM classification

Use the SVM to train the patterns and transformed into its
corresponding feature vector.
Intelligent Database Systems Lab
Methodology (Cont.)


N.Y.U.S.T.
I. M.
Symptom Identification

Word segmentation and Part-Of-Speech (POS) tagging

Semantic dependency graph (SDG) construction.

Semantic label inference.
The identification of symptoms is sentence-based.
t = (modifier , head , relmodifier,head)
Intelligent Database Systems Lab
Methodology (Cont.)
N.Y.U.S.T.
I. M.
P((matters, worry about , goal) | <Anxiety>) is much higher than that in all the other label
Intelligent Database Systems Lab
Methodology (Cont.)

N.Y.U.S.T.
I. M.
Relation Identification

Cause-effect relation

Temporal relation
Intelligent Database Systems Lab
Methodology (Cont.)

Discourse-aware retrieval model

Similarity of events and symptoms
N.Y.U.S.T.
I. M.
Intelligent Database Systems Lab
Methodology (Cont.)
N.Y.U.S.T.
I. M.

Similarity of relations

The relations are represented by symptom chians.

Use the sequence kernel function to calculate the similarity of two
symptom chains.
Intelligent Database Systems Lab
Methodology (Cont.)

N.Y.U.S.T.
I. M.
Sequence kernel function

Symptom 1 ：S1S2S3S4

Symptom 2 ：S3S2S1


Lengths 2 ：{S1S2,S1S3,S1S4,S2S3,S2S4,S3S4} & {S3S2,S3S1,S2S1}
Lengths 3 ：{S1S2S3,S1S2S4,S1S3S4,S2S3S4} & {S3S2S1}
s s ( s1s2 s3 )  s s ( s1s 2 s3 )  1
1 2
2 3
s s (s1s2 s3 )  1
1 3
Intelligent Database Systems Lab
Methodology (Cont.)
N.Y.U.S.T.
I. M.
Intelligent Database Systems Lab
Experiments


N.Y.U.S.T.
I. M.
A total of 3650 consultation documents.

20 documents were randomly selected as the test query set.

100 documents were randomly selected as the tuning set.

The remaining 3530 documents were the reference set to be retrieved.
Use the discounted cumulative gain to evaluate the
retrieval models.

Level 0 : No discourse units are matched.

Level 1 : At least one discourse unit is partially matched.

Level 2 : All of the three discourse units are partially matched.

Level 3 : All of the three discourse units are partially matched, and at
last one discourse unit is exactly matched.
Intelligent Database Systems Lab
Experiments
N.Y.U.S.T.
I. M.
Intelligent Database Systems Lab
Conclusion
N.Y.U.S.T.
I. M.

The discourse information can provide more precise
information about users’ depressive problems.

The psychiatric document retrieval can support
psychological treatments, so people can learn self-help
skills to alleviate their symptoms.

The proposed framework can also be applied to other
domains.
Intelligent Database Systems Lab
Comments

Advantage


The proposed content is easy to know, and the authors use
some instances to explain their ideas.
Drawback


N.Y.U.S.T.
I. M.
…
Application

Psychological document retrieval.

Information Retrieval.
Intelligent Database Systems Lab