Article Structure

Abstract

Two psycholinguistic and psychophysical experiments show that in order to efficiently extract polarity of written texts such as customer-reviews on the Internet, one should concentrate computational efforts on messages in the final position of the text.

Introduction

The ever-growing field of polarity-classification of written texts may benefit greatly from linguistic insights and tools that will allow to efficiently (and thus economically) extract the polarity of written texts, in particular, online customer reviews.

Topic-extraction

One of the basic features required to perform automatic topic-extraction is sentence position.

Polarity-classification vs. Topic-extraction

When dealing with polarity-classification (as with topic-extraction), one should again identify the most aninformative yet dominant proposition of the text.

Method

We aim to show that the last sentence of a customer review is a better predictor for the polarity of the whole review than any other sentence (assuming that the first sentence is devoted to presenting the product or service).

Results

Results of the distribution of differences between the authors’ and the readers’ ratings of the texts are presented in Figure l: The distribution of differences for whole reviews is (un-surprisingly) the narrowest (Figure la).

Conclusions and Future Work

In 2 psycholinguistic and psychophysical experiments, we showed that rating whole cus-tomer-reviews as compared to rating final sentences of these reviews showed an (expected) insignificant difference.