Find the most relevant answer from a set of answer for a particular question. For more information please walthrough this link https://openinnovationgateway.com/ai-nlp-challenge/challenge.php
Note: We executed this project on python3.5 and Also Need to Install difflib library for list comparision
Project Execution Main File is fujitsu_answer_sentence_selection.py
We split this project in two mode: 1) Data generation mode where we create x_train and y_train
from given training data and save it as .txt file in data folder namely x_train.txt, y_train.txt,
x_dev.txt, y_dev.txt. If you want to run this project to create x_train and y_train again
so uncomment create_x_y() and save_x_y() function in fujitsu_answer_sentence_selection.py
2) Second Mode is where we dont need to create x_train and y_train just load them and execute
3) main/evaluation.py is the file to get MRR score as per fujitsu standard
QA_Fujitsu_CNN/
│
├- data/
│ |-- SelQA-ass-result.example.json
│ ├- SelQA-ass-dev.json
│ ├- SelQA-ass-test.json
│ └- SelQA-ass-train.json
| |-- vocab.txt
│ ├- stopwords.txt
│ ├- sentdata.txt
│ └- x_train.txt
| |-- y_train.txt
│ ├- x_dev.txt
│ └- y_dev.txt
│
├- model/
│ │
│ ├- this will contain various trained tensorflow model/
│
├- helper/
│ │
│ ├- data_helper.py
│ |-- dependency_helper.py
│ │-- Evaluation.py
│ │-- FeatureEngineering.py
│ └- pre_processing.py
│
├- main/
│ ├- evaluation.py (FujitSu Standard Evaluation)
│ └- fujitsu_answer_sentence_selection.py
|
└- README.MD