Chatbot Recommendation Engine

Architecture

Two main components type explored with different algorithms,

        1. UserBased Recommendation Engine
            -PearsonCorrelationSimilarity
            -LogLikelihoodSimilarity
            -TanimotoCoefficientSimilarity
            -EuclideanDistanceSimilarity
            -GenericUserSimilarity
            -SpearmanCorrelationSimilarity
        2. ItemBased Recommendation Engine
            -PearsonCorrelationSimilarity
            -LogLikelihoodSimilarity
            -TanimotoCoefficientSimilarity
            -EuclideanDistanceSimilarity

         Evaluated the results with RMSE, F1, Precision and Recall

The REST API is written with Spring

        Endpoint 1:[GET] /getItemBasedRecommendations
        eg:
            http://localhost:8090/getItemBasedRecommendations?userId=200&numberOfRecommendation=6

            output:
            [{"itemID":1,"value":3.5782933},
            {"itemID":19,"value":3.5644608},
            {"itemID":13,"value":3.5610337},
            {"itemID":4,"value":3.5541322},
            {"itemID":17,"value":3.5536952},
            {"itemID":18,"value":3.5515275}]

        Endpoint 2:[GET] /getUserBasedRecommendations
        eg:
           http://localhost:8090/getUserBasedRecommendations?userId=200&numberOfRecommendation=6

           output:
           [{"itemID":1,"value":3.8856046},
           {"itemID":19,"value":3.7924228},
           {"itemID":13,"value":3.5575802},
           {"itemID":18,"value":3.2640123},
           {"itemID":17,"value":3.2016375},
           {"itemID":4,"value":3.1363637}]


        Endpoint 3: [POST] /updateUserData
        eg:
            http://localhost:8090/updateUserData     body: {"userId": "200","itemId": "9","ratings": "5"}

##### Technology 1. Postgrace for loading data 2. Apache.Mahout for recommendations 3. Java Spring Boot for REST api

The spring application is running on port 8090 by default.

Built With

• java