This repository contains various topic analysis services for SingularityNET. The topic analysis methods would include:

• Latent semantic analysis (LSA)
• Probabilistic latent semantic analysis (PLSA)
• Latent Diritchlet allocation (LDA)
• LDA2vec

The services are wrapped using gRPC.

The user provides a collection of documents for topic analysis and the service would return discoverd topics. Each topic consists of a collection of words that represent a given topic.

All services have been tested to work with python3.6.

Please look at the user guide for a detailed spec of the services and how to use the services.

pip install -r requirements.txt

Run the following commands to generate gRPC classes for Python

python3.6 -m grpc_tools.protoc -I. --python_out=.  --grpc_python_out=. service_spec/topic_analysis.proto

python3.6 test_topic_analysis_grpc.py

To start the gRPC server locally

python3.6 topic_analysis_grpc.py

Topic analysis would most likely involve running a dimensionality reduction alogrithm which would take a considerable lenght of time to complete. For this reason, running the service above would return a handle which you will need to use to query a restapi endpoint at a later time. See the user guide for details. You can start that endpoint with the command

python3.6 analysis_results.py

You can also use a suitable application server for python flask. A sample config file using gunicorn is config.py.

Then, you can execute the below command to serve analysis_results.py, by executing gnunicorn in a folder containing config.py, while the configuration file confi.py has a path pointing to analysis_results.py.

gunicorn -c config.py analysis_results:app

LSA:

• Research paper: Indexing by Latent Semantic Analysis
• Wikipedia entry

PLSA:

• Research aper: Unsupervised Learning by Probabilistic Latent Semantic Analysis
• Wikipedia entry

LDA:

• Research paper: Latent Dirichlet Allocation
• Wikipedia entry

LDA2vec:

• Research paper: Mixing Dirichlet Topic Models and Word Embeddings to Make lda2vec

• Eyob Yirdaw

• Eyob Yirdaw