/TextSummarizer

Summarizes texts generating pointer summary

Primary LanguageJupyter Notebook

TextSummarizer

Summarizes texts generating pointer summary

Methodology

  • Tokenize the article into sentences
  • Preprocess the data
    Text Cleaner Class -
    • Tokenize (into sentence and words)
    • Remove noise (stopwords, contractions)
    • Process for noun, verb, adjective and adverb
    • Normalize text (convert to root word)
  • TextRank Algorithm: Similarity matrix using cosine distance
  • PageRank Algorithm
  • Sort sentences in order of importance
  • Generate summary
    Evaluation: Compare with human summary [ROUGE Score]