| dc.description.abstract | Argumentation mining is new field that cuts across many disciplines. Users in Natural Language Processing technologies are keen on presentation of text information in form or arguments due to the many potential applications that are available. For such users, identification of arguments and components that make up the arguments is of much interest since this reduces drastically the effort and resources needed to read through text data that come from many varied sources. There is already attempts to provide applications that are able to read text and identify the arguments in them and the components that make up those arguments. At the moment, tools available are still in research phase. However, use of Argument mining has been incorporated in some domains such as question answering applications and machine translation.
This study proposes the use of machine learning on structure of sentences in order to mine arguments from text. Sentence structure is used in the identification of arguments and later the components that make up such arguments.
In our approach, the machine learning models were created using one data source that contains essays. Evaluation was done on newly acquired data source to as to ascertain the performance of our approach on any new data. Various metrics for evaluation have been presented in this study and discussions on the same done. We have also developed a tool that can be used with any new data. In our tool, we have allowed a user to retrain the model in case of need. The tool provided has a graphical user interface that the user interacts with.
The tool developed in this study, separates the various components of arguments into two distinct categories. | en_US |
