Improving Retrieval of Future-Related Information in Text Collections

Abstract

Abstract-People often want to know expected future events related to given real world entities. For supporting users in the process of future scenario analysis, we propose several methods that enable to retrieve and analyze future-related opinions from large text collections. In particular, we focus on time-unreferenced predictions, which do not contain any explicit future time reference and hence are more difficult to be retrieved. As a second contribution, we propose estimating the validity of predictions by automatically searching for realworld events corresponding to the predictions. This kind of analysis aims to help detect predictions that are no longer valid as well as help estimating prediction accuracy of information sources.

Extracting collective expectations about the future from large text collections

“…Therefore, a future topic/event is extracted only if it is mentioned together with some dates in the future. In fact, news articles also contain many references to future topics and events without mentioning when they are likely to happen [5]. The advantage of considering only time-referenced topics and events is that they allow us to estimate probability distributions over the timeline, and these references are more probable and credible as the time at which they will happen is more or less decided or agreed.…”

“…Time Explorer [8] is a search engine that lets users search in the future and analyze future evolution of topics. Kanazawa et al [5] describe methods for retrieval and validity analysis of future-related information which is not associated with explicit future dates. Our task is different from the above works as we focus on estimating the probability of events based on the aggregated evidence of future-related expressions found in text collections.…”

A Method for Extraction of Future Reference Sentences Based on Semantic Role Labeling

“…In the above examples, the patterns that were matched comprise those studied in previous research [3], [5], [6]. These include time-related expressions ("late May," "from December 1," "from March 1, 1996") and future reference expressions ("is expected," "is planned to," "is likely to").…”

“…Alonso et al [4] have indicated that time information included in a document enhances the effectiveness of information retrieval applications. Kanazawa et al [5] focused on extracting unreferenced future time expressions from a large collection of text, and proposed a method for estimating the validity of the prediction by automatically searching for a real-world event corresponding to the predicted one. Jatowt et al [6] studied the relation between future news written in English, Polish, and Japanese using keywords queried on the web.…”

Multi-lingual Analysis of Future-Related Information on the Web

“…Kanazawa et al [12] estimated that about 30% of predictions in news articles contain future dates based on a small scale investigation. The authors then proposed methods for retrieval and validity analysis of future-related information which is not associated with explicit future dates.…”