Section: New Results

Detecting omissions in journalistic texts

Participants : Héctor Martínez Alonso, Benoît Sagot.

In the journalistic genre that is characteristic of online news, editors make frequent use of citations as prominent information; yet these citations are not always in full. The reasons for leaving information out are often motivated by the political leaning of the news platform.

Existing approaches to the detection of political bias rely on bag-of-words models that examine the words present in the writings. In the context of the VerDI project (see below), we have initiated work aimed at going beyond such approaches, which focus on what is said, by instead focusing on what is ommited. Thus, this method requires a pair of statements; an original one, and a shortened version with some deleted words or spans. The task is then to determine whether the information left out in the second statement conveys substantial additional information. If so, we consider that a certain statement pair presents an omission. To tackle this question, we used a supervised classification framework, for which we require a dataset of sentence pairs, each pair manually annotated for omission.

We have developed a small reference corpus for evaluation purposes, using and comparing both crowd and expert annotation. This corpus has allowed us to examine which features help automaticallly identify cases of omission. In addition to straightforward measures of word overlap (the Dice coefficient), we also determined that there is a good deal of lexical information that determines whether there is an omission. This work is, to the best of our knowledge, the first empirical study on omission identification in statement pairs. We shall make all data and annotations freely available upon publication.