Skip to content

QUICKSTART.md

Latest commit

 

History

History
68 lines (49 loc) · 1.58 KB

QUICKSTART.md

File metadata and controls

68 lines (49 loc) · 1.58 KB
Raw

Quickstart: From LexisNexis DOCX to Research Outputs

This example shows the recommended command-line path for a complete research run.

1. Prepare Environment

python -m venv .venv
.\.venv\Scripts\Activate.ps1
python -m pip install -r requirements.txt
python -m spacy download en_core_web_sm

2. Run the Pipeline

python research_tool.py run -i corpus.docx -o output -t "China; India; Global South" --no-country

Use --no-country when you want a fully offline run. Remove it if you want Wikidata-based country suggestions.

3. Inspect Main Outputs

output/
+-- run_config.json
+-- source_counts.xlsx
+-- merged_sources.xlsx
+-- adjectives_phrases.xlsx
+-- adjectives_final.xlsx
+-- 06_review/
+-- 07_reports/

The most important workbook for discourse analysis is adjectives_phrases.xlsx:

  • KWIC: context for close reading.
  • Collocates: target-window co-occurrence candidates.
  • SemanticProsodyCandidates: candidate appraisal/semantic-prosody labels.
  • GroupComparison: source-group comparison.

4. Generate or Refresh Review Artifacts

python research_tool.py review output --sample-size 50

Fill is_correct, human_result, error_type, and notes in the review templates, then rerun the same command to refresh validation_report.xlsx.

5. Useful Docs

python research_tool.py docs

Start with:

  • docs/WORKFLOW.md
  • docs/METHODOLOGY.md
  • docs/DATA_DICTIONARY.md
  • docs/VALIDATION.md
  • docs/literature_notes/00_INDEX.md

6. Run Tests

python research_tool.py test