Opinion and generic question answering systems

Abstract

The importance of the new textual genres such as blogs or forum entries is growing in parallel with the evolution of the Social Web. This paper presents two corpora of blog posts in English and in Spanish, annotated according to the EmotiBlog annotation scheme. Furthermore, we created 20 factual and opinionated questions for each language and also the Gold Standard for their answers in the corpus. The purpose of our work is to study the challenges involved in a mixed fact and opinion question answering setting by comparing the performance of two Question Answering (QA) systems as far as mixed opinion and factual setting is concerned. The first one is open domain, while the second one is opinionoriented. We evaluate separately the two systems in both languages and propose possible solutions to improve QA systems that have to process mixed questions.

Towards a unified framework for opinion retrieval, mining and summarization

“…This is in line with the research carried out in Balahur et al ((2009aBalahur et al (( ), (2009bBalahur et al (( ),(2010aBalahur et al (( ), (2010b) that have shown that 3-sentence-long snippet retrieval for opinion Table 6 Results of our IR-OM-TS approaches question answering 20 performs better than the traditional one-sentence-long snippet retrieval. This approach is IR p3 -OM A1 -TS, and as far as OM is concerned, the approach dealing with sentiment lexica seems more suitable than the one which uses LSA.…”

“…Concerning the IR component, it is important to mention that a passage length of 1 always obtains poorer results than when it is increased to 3, meaning that the longer the passage, the better. AS we have already mentioned, this conclusion is also supported in (Balahur et al, 2009b), (Balahur et al, 2010b), (Balahur et al, 2010a). When IR is used in conjunction with the TS approach, the results increase for higher compression rates (10%-30%) for the cases where a passage length of 1 is used, in comparison to combining only IR and OM.…”

Analysis and Support of Lifestyle via Emotions Using Social Media

“…These values can be directly compared with the internal emotion q, making it possible to calculate the level of expressiveness ε. Several alternative state-of-the-art approaches on how to extract sentiment from text have recently been discussed in the literature, for example in [2]. Because of the modular nature of the ambient system, one could easily replace a sentiment analysis tool with another solution that measures the sentiment of the message.…”

“…The API provides low latency delivery of public tweets that satisfy the chosen keywords, or chosen hashtags, that identify the specific support domain. 2 The ambient system was implemented using PHP as programming language [29], MySQL as a database [26], and JSON as a data format for storing the values, relating specific users and specific time points in the database [15 ].…”

“…[8]. The main characteristics of Ambient Intelligence applications are that (1) they are nonintrusive, hidden in a person's environment, (2) monitoring makes use of sensor systems, and (3) interventions have a high extent of sensitivity to the context and state of the person. This perspective can be used by not focusing on persons separately but, as is done in this paper, on the social interactions they have.…”