Extract Terms From a Document

Follow these steps to extract term candidates from a monolingual document into a glossary.

About this task

Term extraction scans a document and proposes frequently occurring word sequences as candidate terms, which are reviewed before being saved to a glossary. Common words are filtered out automatically using stop words. For background, see Term Extraction.

Procedure

  1. Select TasksTerm Extraction from the main menu or click the Term Extraction button on the toolbar.
    The Term Extraction dialog is displayed:
    Term Extraction Dialog
  2. Type the name of the document to analyze in the Source File box or click the Browse... button next to it to select it from the file system.
  3. Type the name of the glossary to create in the GlossML File box or click the Browse... button next to it to select a name and location.
  4. Select the language of the document from the Source Language drop-down list.
  5. If necessary, correct the format in the File Type drop-down list.
    Anchovy uses OpenXLIFF to detect the file type and character set automatically, so they usually do not need to be changed. Manual selection is only needed when the format is ambiguous, as with plain text files. The list contains the document formats supported by OpenXLIFF, such as office documents, HTML and XML files.
  6. If necessary, correct the character set in the Character Set drop-down list.
  7. Indicate in the Minimum Frequency selector the minimum number of times a candidate term must appear in the document.
    A higher value keeps only terms that appear often, producing fewer but more reliable candidates. A lower value surfaces more candidates, including infrequent ones.
  8. Indicate in the Maximum Term Length selector the maximum number of words a candidate term can contain.
    Term length is measured in words. Set it to 1 to extract only single words, or to a higher value to also capture multi-word phrases.
  9. Click the Extract Candidates button.
    The document is analyzed and the list of candidate terms is displayed in the Term Candidates dialog:
    Term Candidates Dialog
  10. Review the list of candidates. Select the candidates to discard and click the Remove Selected button.
    Besides the candidate term, the list shows a Score and the values used to calculate it: Casing, Position, Frequency, Relevance, Relatedness and Different. These are statistically relevant for deciding what to propose, but they do not need to be interpreted. Read the candidate terms and keep or discard each one using professional judgement.
  11. Optionally, select the Open generated glossary checkbox to open the new glossary in Anchovy after saving.
  12. Click the Save Glossary button to store the remaining candidates in the glossary.

Results

A glossary containing the selected term candidates is created and, optionally, opened in Anchovy.