2019/10/26

First Phase - 80%

  1. Front-end program can extract sentence from speech with high accuracy
  2. Currently no parameters or specific actions can be determined
Python speech to text → Part-Of-Speech tagger

TODO

  1. Extract of device name (NOUN type)
  2. Start of second phase

2019/11/2

First Phase - 100%

  1. Actions can be determined
  2. Back-end analysis can extract DEVICE NOUN from text

TODO

  1. Continue onto second phase

2019/11/9

Second Phase – 40%

  1. Added Syntatic Dependency Parsing allowing creation of parse trees
  2. Separation of DEVICE NOUN, PARAMETER NOUN and their corresponding VERB
  3. Elimination of stopwords
Python speech to text → Part-Of-Speech tagger → Syntatic Dependency Parser → Rule Matching

TODO

  1. Rule-based matching to account for inaccuracy in parse trees
  2. Word embeddings(BERT/Flair) for synonyms

2019/11/16

Second Phase – 60%

  1. Started revising output format to a 4 element tuple
  2. Research about how to handle if statement

TODO

  1. Finish the 4 element tuple
  2. Re-dig into SpaCy for different command structure

2019/11/23

Second Phase – 85%

  1. Synonym partially done
  2. New output format basically done with some minor errors to deal with

TODO

  1. Add robustness of strcture analysis
  2. Wrap off handling synonym & acronym

2019/11/30

Second Phase – 100%

  1. Synonym & acronym handling done
  2. Robustness enhanced

TODO

  1. Start doing research for the stretch goal
  2. Extra effort on system robustness
  3. Handle if statement

2019/12/7

Third Phase Head Start

  1. Research about third phase
  2. If statement partially done

TODO

  1. Final presentation preparation
  2. Finish if statement
  3. Weblite update
  4. Final video