Abstract Online reputation management (ORM) comprises the collection of techniques that help monitoring and improving the public image of an entity (companies, products, institutions) on the Internet. The ORM experts try to minimize the negative impact of the information about an entity while maximizing the positive material for being more trustworthy to the customers. Due to the huge amount of information that is published on the Internet every day, there is a need to summarize the entire flow of information to obtain only those data that are relevant to the entities. Traditionally the automatic summarization task in the ORM scenario takes some in‐domain signals into account such as popularity, polarity for reputation and novelty but exists other feature to be considered, the authority of the people. This authority depends on the ability to convince others and therefore to influence opinions. In this work, we propose the use of authority signals that measures the influence of a user jointly with (a) priority signals related to the ORM domain and (b) information regarding the different topics that influential people is talking about. Our results indicate that the use of authority signals may significantly improve the quality of the summaries that are automatically generated.

Given the task of finding influencers (opinion makers) for a given domain in a social network, we investigate (a) what is the relative importance of domain and authority signals, (b) what is the most effective way of combining signals (voting, classification, learning to rank, etc.) and how best to model the vocabulary signal, and (c) how large is the gap between supervised and unsupervised methods and what are the practical consequences. Our best results on the RepLab dataset (which improves the state of the art) uses language models to learn the domain‐specific vocabulary used by influencers and combines domain and authority models using a Learning to Rank algorithm. Our experiments show that (a) both authority and domain evidence can be trained from the vocabulary of influencers; (b) once the language of influencers is modeled as a likelihood signal, further supervised learning and additional network‐based signals only provide marginal improvements; and (c) the availability of training data sets is crucial to obtain competitive results in the task. Our most remarkable finding is that influencers do use a distinctive vocabulary, which is a more reliable signal than nontextual network indicators such as the number of followers, retweets, and so on.

In this article, we investigate what sorts of information humans request about geographical objects of the same type. For example, Edinburgh Castle and Bodiam Castle are two objects of the same type: “castle.” The question is whether specific information is requested for the object type “castle” and how this information differs for objects of other types (e.g., church, museum, or lake). We aim to answer this question using an online survey. In the survey, we showed 184 participants 200 images pertaining to urban and rural objects and asked them to write questions for which they would like to know the answers when seeing those objects. Our analysis of the 6,169 questions collected in the survey shows that humans have shared ideas of what to ask about geographical objects. When the object types resemble each other (e.g., church and temple), the requested information is similar for the objects of these types. Otherwise, the information is specific to an object type. Our results may be very useful in guiding Natural Language Processing tasks involving automatic generation of templates for image descriptions and their assessment, as well as image indexing and organization.

Negation, intensifiers, and modality are common linguistic constructions that may modify the emotional meaning of the text and therefore need to be taken into consideration in sentiment analysis. Negation is usually considered as a polarity shifter, whereas intensifiers are regarded as amplifiers or diminishers of the strength of such polarity. Modality, in turn, has only been addressed in a very naïve fashion, so that modal forms are treated as polarity blockers. However, processing these constructions as mere polarity modifiers may be adequate for polarity classification, but it is not enough for more complex tasks (e.g., intensity classification), for which a more fine‐grained model based on emotions is needed. In this work, we study the effect of modifiers on the emotions affected by them and propose a model of negation, intensifiers, and modality especially conceived for sentiment analysis tasks. We compare our emotion‐based strategy with two traditional approaches based on polar expressions and find that representing the text as a set of emotions increases accuracy in different classification tasks and that this representation allows for a more accurate modeling of modifiers that results in further classification improvements. We also study the most common uses of modifiers in opinionated texts and quantify their impact in polarity and intensity classification. Finally, we analyze the joint effect of emotional modifiers and find that interesting synergies exist between them.