Automatic Text Summarization for Amharic Legal Judgments
No Thumbnail Available
Date
2006-09
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Addis Ababa University
Abstract
With the continuing fast growth of information and data today, it has become
more and mo re urgent and important to find proper information efficiently, with
some improved mechanisms
Nowadays, people need much more information in work and life. The use of
information techno logy such as the Internet makes information more easily
accessible. However, people are having problems to easily get the information
they want in a summarized way without wasting their time in the vast load of
information made available to them . Thus, automatic text summarization draws
substantial interest s incest provides a solution to the information overload
problem people face in this digit al era.
This work is concentrated on producing a prototype system of text summarization
on Amharic legal judgments. The methodology employed is an extraction
technique.
Amharic legal judgments rendered by the supreme court of Ethiopia are selected
in consultation with legal experts (lawyers , law instructor & students). The data
selected in this manner are employed to generate the summaries
Sentences in each judgment are classified in to different unit s according to their
argumentative roles. From each argumentative unit , sentences with the highest
weight at 20 % compression rate are extracted and presented as a summary
. To evaluate the performance of the system, a random summary at similar
compression rate is generated. Using extrinsic evaluation technique, the
performance of the system summary and the random summary were compared
with an ideal summary (human generated summary).
The results obtained from the system are promising when compared with the
random summary.
To improve the performance of the system, sentences at 10% compression rate
were extracted and the system's performance has improved. Therefore, the
sentences extracted by the system summary using different extraction features
are much closer to the manual (ideal) summary.
The prototype text summarize has been developed using the Python
programming language as a tool.
Description
Keywords
Information Science