Document Summarization - Elias-Martha - Final

Post on 08-Apr-2018

220 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

Transcript

8/6/2019 Document Summarization - Elias-Martha - Final

http://slidepdf.com/reader/full/document-summarization-elias-martha-final 1/27

Document Summarization

Elias Haileselassie

Martha Teffera

May 14, 2011

School of Information Science

Addis Ababa University

8/6/2019 Document Summarization - Elias-Martha - Final

http://slidepdf.com/reader/full/document-summarization-elias-martha-final 2/27

Outline

 What is Summarization

Input & Output

Examples

Types of Summaries

Single document vs multi document

Extract vs Abstract

Generic vs Query Oriented Functions

Indicative

Informative

Summarization Techniques Sentence Extraction

Evaluating Summaries

Applications

Pros and Cons

8/6/2019 Document Summarization - Elias-Martha - Final

http://slidepdf.com/reader/full/document-summarization-elias-martha-final 3/27

What is Summarization?

Summarization is the creation of ashortened version of a documenteither manually or automatically 

8/6/2019 Document Summarization - Elias-Martha - Final

http://slidepdf.com/reader/full/document-summarization-elias-martha-final 4/27

What is Summarization? …cnt’d

 Automatic Document Summarization is the

creation of a shortened version (summary) of adocument by a computer program with outlosing content

8/6/2019 Document Summarization - Elias-Martha - Final

http://slidepdf.com/reader/full/document-summarization-elias-martha-final 5/27

Input & Output

A summary input can be in the form of apicture, a movie, an audio segmentor a text document

The output to be summarized may be inthese different multimedia forms.

8/6/2019 Document Summarization - Elias-Martha - Final

http://slidepdf.com/reader/full/document-summarization-elias-martha-final 6/27

Input & Output …cnt’d

Input

Text

Compression Rate or Summary Length

rate = Summary Length/Source Length

Output

Summary Text/Audio

8/6/2019 Document Summarization - Elias-Martha - Final

http://slidepdf.com/reader/full/document-summarization-elias-martha-final 7/27

Examples of Summaries Newspaper headlines

A preview of a video show

Abstracts of scientific articles Reviews of books and movies

A library catalog of abstracts of articles

A table of contents for a book ,magazine or any kind of document

8/6/2019 Document Summarization - Elias-Martha - Final

http://slidepdf.com/reader/full/document-summarization-elias-martha-final 8/27

Types of summarization

Depending on the input

Single document summaries  Multi document summaries

8/6/2019 Document Summarization - Elias-Martha - Final

http://slidepdf.com/reader/full/document-summarization-elias-martha-final 9/27

Types of summarization …cnt’d

Depending on the output(process)Extract summaries

 Abstract summaries

8/6/2019 Document Summarization - Elias-Martha - Final

http://slidepdf.com/reader/full/document-summarization-elias-martha-final 10/27

 An extract is a summary consisting entirely of materialcopied from the input.

It is a custom to speak of an extract of K%condensation: so K% of the input's words/sentences/paragraphs may appear in theextract.

Extract Summaries

Types of summarization …cnt’d

8/6/2019 Document Summarization - Elias-Martha - Final

http://slidepdf.com/reader/full/document-summarization-elias-martha-final 11/27

In contrast, an abstract is a summary at least some of  whose material is not present in the input

In general, abstracts offer the possibility of higherdegrees of condensation: a short abstract may offermore information than a longer extract.

 Abstract Summaries

Types of summarization …cnt’d

8/6/2019 Document Summarization - Elias-Martha - Final

http://slidepdf.com/reader/full/document-summarization-elias-martha-final 12/27

Generic vs Query-OrientedDepending on the purpose, a summary can be

Generic - reflect the author's point of view withrespect to all important topics in the input text

Query Oriented (user-focused ) - reflect only the topics

in the input text that are specific to a given query 

Types of summarization …cnt’d

8/6/2019 Document Summarization - Elias-Martha - Final

http://slidepdf.com/reader/full/document-summarization-elias-martha-final 13/27

Indicating what is inside a document

helping the user to decide whether to read theinformation source, or not

Replace the original document with out losing content

Saves time

Functions of summarization

8/6/2019 Document Summarization - Elias-Martha - Final

http://slidepdf.com/reader/full/document-summarization-elias-martha-final 14/27

Indicative summaries An indicative summary can provide only an indication

of the main topics in the input text

Thus, an indicative abstract is aimed at helping theuser to decide whether to read the informationsource, or not

The main purpose is to suggest the contents of the

document without giving away detail on thedocument’s content.

Functions of summarization

8/6/2019 Document Summarization - Elias-Martha - Final

http://slidepdf.com/reader/full/document-summarization-elias-martha-final 15/27

Informative summaries

 An informative summary is meant to represent (and

often replace) the original document. It should contain all the pertinent information

necessary to convey the core information and omitunnecessary information

Functions of summarization …cnt’d

8/6/2019 Document Summarization - Elias-Martha - Final

http://slidepdf.com/reader/full/document-summarization-elias-martha-final 16/27

 An ideal method

1. Understanding the text

2. Transforming the analytical result of the text into arepresentation(model) of a summary 

3. Generating a summary from the internal

representation

Summarization Techniques

8/6/2019 Document Summarization - Elias-Martha - Final

http://slidepdf.com/reader/full/document-summarization-elias-martha-final 17/27

Current methods

1. Extracting the important parts (sentences) in the text

2. Just arranging them in the order in the text

Summarization Techniques cnt’d

8/6/2019 Document Summarization - Elias-Martha - Final

http://slidepdf.com/reader/full/document-summarization-elias-martha-final 18/27

Sentence Extraction

   Word frequency in text

  Title of text  Position of sentence

  Cue phrases in text

Summarization Techniques cnt’d

8/6/2019 Document Summarization - Elias-Martha - Final

http://slidepdf.com/reader/full/document-summarization-elias-martha-final 19/27

 Ideal evaluation

Measure the information content

Compression Ratio: CR = (length S) / (length doc)

Retention Ratio: RR = (info in S) / (info in T)

Evaluating Summaries

l

8/6/2019 Document Summarization - Elias-Martha - Final

http://slidepdf.com/reader/full/document-summarization-elias-martha-final 20/27

 Practical evaluation

Comparing with existing summaries for similarity   word overlap

  Measure Recall & Precision -> 2PR/(P+R)

Evaluating Summaries …cnt’d

Relevant Non Relevant

System Relevant A B

System Non-Relev. C D

 B A

 A P

+

=:Precision

C  A

 A  R

+

=:Recall

Applications

8/6/2019 Document Summarization - Elias-Martha - Final

http://slidepdf.com/reader/full/document-summarization-elias-martha-final 21/27

ApplicationsSearch Engines: Result Summarization

8/6/2019 Document Summarization - Elias-Martha - Final

http://slidepdf.com/reader/full/document-summarization-elias-martha-final 22/27

Applications …cnt’d

shablast.com

8/6/2019 Document Summarization - Elias-Martha - Final

http://slidepdf.com/reader/full/document-summarization-elias-martha-final 23/27

Applications …cnt’d

Table of Contents Generation - MS Word

8/6/2019 Document Summarization - Elias-Martha - Final

http://slidepdf.com/reader/full/document-summarization-elias-martha-final 24/27

Applications …cnt’d

News Headline/Summary Generation

8/6/2019 Document Summarization - Elias-Martha - Final

http://slidepdf.com/reader/full/document-summarization-elias-martha-final 25/27

Applications …cnt’d

Book/Article Summarization - Copernic Summarizer

8/6/2019 Document Summarization - Elias-Martha - Final

http://slidepdf.com/reader/full/document-summarization-elias-martha-final 26/27

Pros and Cons

ProsSaves time

Helps decision making/to read or

not to readHelps select the right

document(information overload)

Cons

Slows search engine result display

8/6/2019 Document Summarization - Elias-Martha - Final

http://slidepdf.com/reader/full/document-summarization-elias-martha-final 27/27

Thank you!

top related