Improving Out-of-Domain Sentiment Polarity Classification Using Argumentation

Abstract

Abstract-Domain dependence is an issue that most researchers in corpus-based computational linguistics have faced at one time or another. With this paper we describe a method to perform sentiment polarity classification across domains that utilises Argumentation. We train standard supervised classifiers on a corpus and then attempt to classify instances from a separate corpus, whose contents are concerned with different domains (e.g. sentences from film reviews vs. Tweets). As expected the classifiers perform poorly and we improve upon the use of a simple classifier for out-of-domain classification by taking class labels suggested by classifiers and arguing about their validity. Whenever we can find enough arguments suggesting a mistake has been made by the classifier we change the class label according to what the arguments tell us. By arguing about class labels we are able to improve F 1 measures by as much as 14 points, with an average improvement of F 1 = 7.33 across all experiments.

Abduction and Argumentation for Explainable Machine Learning: A Position Survey

“…(ii) priorities are determined by the learning data [11,38,39,40,50,54,60]; or (iii) priorities are unavailable or capture domain expert knowledge and are provided externally [3,5,6,17,18,19,41,42,48]. Additional considerations regarding a subset of the works reviewed below can be found in [7].…”

“…Classification enhanced with Argumentation (CleAr) [5,6] considers a setting where expert opinion or domain knowledge on how a new input should be classified might be available, and these alternatives need to compete against or support the learned hypothesis and each other. Argumentation helps resolve the conflicts that arise, with each argument being assigned a base score, and these scores being used to decide which conclusion prevails.…”

Detecting deceptive reviews using Argumentation

“…Thus, these argumentation features can be seen as adding a semantic layer to the analysis of deceptive behaviour in reviews on top of the syntactic analysis given by standard NLP when using machine learning techniques. Our approach can also be seen as integrating argumentation and machine learning, in the spirit, for instance, of [28,19,9,8], but in a different context (deception detection) and using a novel methodology (argumentative features).…”