In this bot we use three knowledge retrievers. In this folder, we implement the knowledge based not the retriver function (e.g., tfidf).
We use the Wiki dump and mongo DB provided by KILT. Please follow the instruction in https://github.com/facebookresearch/KILT#kilt-knowledge-source. Once it is setup, then you can run:
sudo mongod --bind_ip_all
For the interactive script, we use the wikipedia title retriever from https://github.com/goldsmith/Wikipedia.
We use the code in https://github.com/JulesGM/ParlAI_SearchEngine. Follow the readme of the page to set up the search engine. Then you can run:
python search_server.py serve --host 0.0.0.0:8081 --search_engine="Bing" --use_description_only --subscription_key "YOUR_KEY"
or, if you don't have a bing key:
python search_server.py serve --host 0.0.0.0:8081
which uses google search.
As in AdapterBot, we use Neo4J to setup the KG. Please follow the instruction in https://github.com/HLTCHKUST/adapterbot/tree/main/retriever/graphdb to set up the KG and run:
sudo service neo4j start