Here, is a short description of the contributions of the WSE group to the Qanary ecosystem. It consists of several artifacts that are used to build Knowledge Base Question Answering systems. All are located in their own GitHub repositories.
-
Qanary is a framework for building Question Answering systems built on top of the Spring Boot framework and using RDF annotations to represent the knowledge about a given user’s question. The basic functionality of the Qanary is to provide the orchestration of Qanary components.
-
Build and run the complete Qanary Question Answering Pipeline in a Docker container.
-
Create a Java component from the Qanary Maven Archetype.
-
Create a Python component using the Qanary Python Helper Library.
-
The Qanary components implement tasks that are needed to answer questions in particular domains. Typically, each component solves one task only (e.g., Language Detection, Named Entity Recognition, Relation Detection, Expected Answer Type Classification, Query Building, Answer Candidate Filtering).
-
For establishing the data exchange between components, the Question Answering vocabulary (short: qa) is used.
-
A complete, ready-to-run minimal example of 3 Qanary-driven Python components (and the corresponding orchestrator as well as the required triplestore) is prepared as a Docker compose script.
-
The Qanary Chatbot UI is providing an easy-to-use web-based chatbot frontend that can be connected to the Qanary framework APIs.
