Recurrent Point Review Models

This is a companion code for the paper "Recurrent Point Review Models"

Description

Deep neural network models represent the state-of-the-art methodologies for natural language processing. Here we build on top of these methodologies to incorporate temporal information and model how review data changes over time. We use the dynamic representations of recurrent point process models, which encode the history of how reviews of businesses and services arrive in time, to generate instantaneous language models with improved prediction capabilities. Simultaneously,our methodologies improve the predictive power of our point process models by incorporating summarized review content representations. We provide recurrent network and temporal convolution solutions for modeling the review content. We deploy our methodologies in the context of recommender systems,effectively characterizing the change in preference and taste of users as time evolves..

Note

For the purpose of the reproducibility of the results for this paper we have created self containing package with detailed description of how to run the models.

Data uploading and pre-processing

For storing the data we use MongoDB, therefore it is pre-request for running the code to have installed MongoDB. Detailed explanation of how to install the MongoDB can be found on the official web page.

Once you have installed mongoDB please do the following steps:

1. Download the Yelp 2019 dataset and copy the data into scripts/preprocessing/yelp/ folder.
2. run the sh upload.sh script.

Installation of the dpp library

In order to run the code please install dpp library.

Install Tyche

1. cd Tyche
2. pip install -r requirements.txt
3. pip install .

Install GENTEXT

1. cd GENTEXT
2. pip install -r requirements.txt
3. pip install .

Install dpp

1. cd deep_point_process
2. pip install -r requirements.txt
3. pip install .

Running training scripts

For running the training please use the train_script.py and choose one of the experiments.

1. cd deep_point_process
2. python scripts/train_script.py experiments/business_text_pp_cnn_emb16.yaml

Running inference scripts

For running the inference please use the inference_script.py and choose one of the experiments.

1. cd deep_point_process
2. python scripts/inference_script.py models/yelp19_shopping_business_text_pp_cnn_model_tpp_model_embedding_size_64_model_tpp_model_cell_type_hidden_size_16