Pickaxe

The purpose of this project was to develop a search engine which would be capable of ranking and retrieving arguments based on relevance to certain keywords. These argument were extracted from text sources such articles and blog posts through Machine Learning and Natural Language Processing tools. These arguments would then be structured within a machine readable format which would be in XML or JSON.

The main objective was to establish an advanced search mechanism tailored for arguments, leveraging Elasticsearch's capabilities for argument ranking and retrieval based on keyword relevance. Additionally, the search engine featured functionality to manipulate argument data and present it visually through argument maps to display how premises and conclusion would be linked together. Arguments could also be displayed as being "for" or "against" certain topics or discussion points.

The project also incorporated a Flask application, which served as the user-facing interface for the argument search engine. This application functioned as the gateway for users to input search terms and retrieve indexed and ranked arguments thanks to the Elasticsearch integration.

Argument data would be stored within the SADFace format. Due to a lack of argument data available in this format and SADFace being relatively new compared to the AIF format, a script was developed with the aim of converting arguments from the AIF format to SADFace. This increases the availability of data resources for the project.