/ineuron-ml-hackathon

Provides solutions to students’ queries on skype on a daily basis for the past 2 years. Almost in each batch students have repetitive queries which can be solved using some advanced virtual assistance. Build an intelligent virtual assistant to solve duplicate queries. Data Gathering: Use skype chat data to create an intelligent virtual assistant. Bot in such a way that it can use historical chat data to solve a query that was already solved earlier.

Primary LanguagePython

drawing

KIRA

A smart bot trained using Ineuron twitter dataset.





Preprocessing the data :

We process the given data by extracting the text from json file , and pairing the questions with answers. In the end we got more 4 lakh question answer pair sets. We used this data for our prediction of answer to repetitive questions.

drawing

Processing the data :

We have used NLTK library to process our data .

  • We dropped duplicated questions set.
  • We dropped null questions set.
  • We removed stopwords and lemmatized the question text.
  • We tokenize the text.
  • By using above steps we transformed the data.

User input :

  • User will enter a question
  • The question will be transformed by using above steps
  • The tokenized question will be compared with saved data and mathcing answer will be printed out.
  • The question which is similar to asked question will be printed out