Automatic Eurovoc indexing an Experiment in the Czech Parliament

1
Automatic Eurovoc indexing an Experiment in the
Czech Parliament

• Anna Lhotská, Václav Sklenár
• Office of the Chamber of Deputies, Parliament of
  the Czech Republic

2
History of the Czech version

• 1993 - preliminary translation of the second
  version into Czech
• 1995 - Czech version - edition 3.0
• 07/2003 - Czech version - edition 4.0.
• 07/2004 - Czech version edition 4.1
• The Czech Eurovoc is fully compatible with other
  official language versions

3
Application of Eurovoc in the Information System
of Parliament

• library database - aRL
• database of petitions - Lotus Notes
• intellectual indexing of parliamentary documents
• multilingual searching in parliamentary
  documentation

4
Manual indexing/1

• all parliamentary documents that are publicly
  accessible in full text in an electronic form via
  Internet are indexed intellectually with Eurovoc
  terms. At present it represents 3.500 documents
• retrospective indexing of older materials
  continues, great number of older documents still
  remains not indexed

5
Manual indexing/2

• document types - bills, budgets, agreements,
  agendas, parliamentary questions
• classification - 127 Eurovoc Microthesaury
• indexing - descriptors from Eurovoc -
  multilingual searching
• indexing - descriptors from complementary
  thesaurus - searching in Czech language only

6
Automatic indexing tool/1

• SELECTION of terms from document text
• STOP-WORDS LIST (negative dictionary)
• LEMMATIZER - set of rules for grammatical
  alterations
• COMPARISION of a set of basic-form words from a
  text with Eurovoc terms

7
Automatic indexing tool/2

• MULTIWORD EXPRESSIONS recognition
• NON-DESCRIPTOR/DESCRIPTOR transformation
• ABSOLUTE FREQUENCY WEIGHTING

8
Limitations of the automatic indexing tool

• insufficient term weighting system
• term location within the text (non-structured
  texts)
• insufficient recognition of multiword expressions
• lack of automatic non-descriptor/descriptor
  proposals.

9
Thank you for your attentionAnna Lhotská,
lhotska_at_psp.czVáclav Sklenár, sklenar_at_psp.cz