This is the machine learning micro-service, which is responsible for re-ranking places based on their popularity on Twitter

• accept POST requests from the web clients side (query, user location, and places)
• the function build_dataset() is called with the following parameteres:
  • documents : JSON that contains the places resulting from the elastic search
  • user_location : [longitude, latitude]
  • query : the inputed query
  • collection: tweets collection
• the rereank() function is called with the following parameters:
  • loaded_model: the machine learning model (loaded from main)
  • data : the dataset returned from build_dataset() this function takes the dataset and reranks the documents and finally returns the results in json format
• return re-ranked places

• run the command docker build -t ml_service:latest
  • the -t is used to set the a TAG for the docker image

• run the command python ml_service.py

• send a POST request to the service containing query, user location, and places
• example of the passed data:

    { 
      "query" : "starbucks",
      "lat" : 40.224,
      "lon" : -70.345,
      "places" : [
                    {"_source":{
                            "location" : { "lat" : 41.35, "lon" : -71},
                            "name" : "starbucks"
                              }}
                              ....
                  ]
    }