darya-makarenko/intent_classification

Intent Classification Service

The service uses SpaCy and nltk to preprocess text and get it to normalised form without punctuation and then convert it to vector representation.

Docker usage:

1. To build docker container type docker build -t intent_classify . in intent_classify folder.
2. To run Docker container type docker run -p 5000:5000 intent_classify.

Service API:

1. To start Swagger and get all available methods visit <base_url>:5000. Example: 127.0.0.1:5000.
2. If you want to get information about the intent classification service, make GET request to the endpoint <base_url>:5000/classify_intent. NB: You can perform all requests with the help of Swagger (see par. 1), or you can choose Postman, curl or your custom web-client.
3. In order to classify one or more texts send POST request to the endpoint <base_url>:5000/classify_intent with JSON object in the request body:

  {
  "texts": [
      {
        "text": "нет, не правильно",
        "lang": "ru"
      }
  ]}
  

Project details:

1. main.py - start point of the project.
2. view.py - specifies service endpoints.
3. intent_classify.py - implements intent classification functionality.
4. positive.csv - dataset of positive (yes) phrases.
5. negative.csv - dataset of negative (no) phrases.

methods of "intent_classify.py"

• def cleanup_text(text, logging=False): deletes punctuation from texts and normalises them
• def classify_intent(texts): expects an array of texts and produces the array of following structure:

      [
          {
            'text': text['text'], 
            'lang': text['lang'],
            'intents': [
              { 'label': 'pos',
                'conf': preds[0][0]
              },
              {'label': 'neg', 
               'conf': preds[0][1]
              }]
          }
      ]

Docker problems:

1. If you get problem with already used ports when starting docker image type: docker ps to show list of running containers and kill (see par. 4) container that uses this port.
2. If first advice didn't work, try typing docker container prune.
3. You can then try typing docker system prune.
4. To kill running docker container type docker ps to show list of running containers, then use command kill <deleted_container_id>. You can either use id of running container or it's name.
5. To see list of all built docker images type docker images.