Release note of Text Structure Analysis System

[1] Overview

Texts such as editorials are written to show writer's opinion and thinking
to readers.  We call such opinion "Writer's Intention".  We have developed a
system which analyzes the rhetorical structure of Japanese texts from the
view point of the Writer's Intention.  This system makes use of five kinds
of sentence information obtained from the results of morphological analysis:
sentence type which is classified by its intention, topic information,
demonstrative pronouns, conjunctive phrases and repeated words, and it
analyzes a text structure using about 400 rules.  These rules are based on
analysis of Japanese context phenomena and are divided into 4 categories.
The first category is for analysis of the relationship between a sentence
and its adjacent one.  The second category generates some groups of
sentences using adjacent sentence relationships.  The relationship between
these groups will be analyzed in the third category.  The fourth category
develops text structure as the result of recognition of their elements.

[2] System environment

This system is available on SIMPOS ver.7.1 of PSI system.

[3] File list
  README		Readme file
  README.j (K)		Readme file (in Japanese)
  INSTALL		Installation guide
  INSTALL.j (K)		Installation guide (in Japanese)
  ctx_system.esp (K)	Source file (main module)
  make_txt.esp (K)	Source file (text conversion module)
  analyz_sty.esp (K)	Source file (sentence analysis module)
  initial_wm.esp (K)	Source file (working memory initialization module)
  make_wmdat.esp (K)	Source file (working memory generation module)
  msage_win.esp		Source file (message window module)
  str_dsp.esp (K)	Source file (text structure display module)
  str_dsp_ch.esp (K)	Source file (text structure display sub-module)
  lex.esp		Source file (rule compiler: morphlogical analyzer)
  parse.esp		Source file (rule compiler: syntax analyzer)
  codegen.esp		Source file (rule compiler: code generator)
  struct_ext.esp (K)	Source file (external function for rules)
  mor.tmp		Template file for morphological analysis data
  prs.tmp		Template file for syntax analysis data
  code.1		Code generation file
  out_ph12.esp (K)	Compiled rules (phase1,2)
  out_ph13.esp (K)	Compiled rules (phase1,3)
  out_ph14.esp (K)	Compiled rules (phase1,4)
  out_ph5.esp (K)	Compiled rules (phase5)
  out_ph6.esp (K)	Compiled rules (phase6)
  txt_struct.rule (K)	Context structure analysis rule
  1002_2.org (K)	Example (original text)
  1002_2.txt (K)	Example (formated text)
  1002_2.mor (K)	Example (morphological analysis text)
  1002_2.init (K)	Example (output of sentence analysis)
  (Files that have the mark '(K)' on their right include Kanji letters.)

[4] Install
 (1) Make the following directories.
        init		(for *.init files)
        org		(for *.org files)
        morph		(for *.mor files)
        rule		(for *.rule and code.1 files)
        source		(for *.esp, *.tmp files)
        text		(for *.txt files)
        tmp		(for error files)

 (2) Catalogue *.tmp files.
	mor.tmp		prs.tmp

 (3) Catalogue *.esp files.
	analyz_sty.esp	lex.esp		out_ph12.esp	out_ph6.esp
	codegen.esp	make_txt.esp	out_ph13.esp	parse.esp
	ctx_system.esp	make_wmdat.esp	out_ph14.esp	str_dsp.esp
	initial_wm.esp	msage_win.esp	out_ph5.esp	str_dsp_ch.esp
	struct_ext.esp
  (The details are shown in INSTALL/INSTALL.j.)

[5] References
  (1) Fukumoto, J. : "Context Structure Analysis based on the Writer's
    Insistence", IPSJ SIG-NL 78-15, pp.113-120, 1990. (in Japanese)
  (2) Fukumoto, J. and Yasuhara, H. : "Japanese Text Structure Analysis",
    IPSJ SIG-NL 85-11, pp.81-88, 1991. (in Japanese)
  (3) Fukumoto, J. and Yasuhara, H. : "Text Structure Analysis based on
    Sentence Cohesion", IPSJ SIG-NL 88-2, pp.9-16, 1992. (in Japanese)
