Predict Service Based On RNN-LSTM And User Interest Information Of Voice Interface

Ji-Hoon Lee · Nammee Moon — first author·KIBME Fall Conference 2019 (Undergraduate)2019Conference

PDF ↓

As people hold more devices, filtering information down to what suits them gets harder. The voice interface recommends search terms based on the interest information given at sign-up.

  • Filters crawled information and recommends search terms using gender, age and residence
  • When the interest is finance or stocks, an RNN-family LSTM predicts the stock price
  • Prediction was tested against Apple’s stock price