Search

Information is powerful; it can influence peoples' views and, in turn, their behavior. Much recent research and discussion on the role information plays in view change has focused on filter bubbles, echo chambers and misinformation and how they might influence what people think and how they act. However, no prior work has focused specifically on understanding the human information behavior (HIB) that drives and facilitates view change. We report findings from interviews with 18 people who recently changed views on issues they considered important. We found a tight symbiotic relationship between HIB and view change; passive information encountering sparked change, often spurring follow-up active seeking and verification which progressed the change to a “point of no return,” supported making the change and reinforced the decision to change. When shared, information that contributed to the change sometimes sparked changes in others (as did expressing or debating the change), serving as an information encounter that perpetuated a cycle of HIB and view change. This understanding of the integral role of HIB in view change can inform policy and systems design to promote view change autonomy and a broader research agenda of understanding HIB to support democratic principles and values.

There has been considerable hype about filter bubbles and echo chambers influencing the views of information consumers. The fear is that these technologies are undermining democracy by swaying opinion and creating an uninformed, polarised populace. The literature in this space is mostly techno-centric, addressing the impact of technology. In contrast, our work is the first research in the information interaction field to examine changing viewpoints from a human-centric perspective. It provides a new understanding of view change and how we might support informed, autonomous view change behaviour. We interviewed 18 participants about a self-identified change of view, and the information touchpoints they engaged with along the way. In this paper we present the information types and sources that informed changes of viewpoint, and the ways in which our participants interacted with that information. We describe our findings in the context of the techno-centric literature and suggest principles for designing digital information environments that support user autonomy and reflection in viewpoint formation.

In a ubiquitous computing scenario, characterized by pervasive technologies, tourists can get assistance from mobile technologies in planning their trips. In a context where more and more people own smartphones, tourists expect to get personalized suggestions just in time whenever and wherever they need. To be effective, mobile applications for travel recommendation should consider both the variability of the user’s interests and an effective way to express them while interacting with the environment. This paper presents LOOKER, a mobile recommender system for tourism and travel-related services that considers the above-described issues. It is an adaptable application developed for the Android platform, which takes into account basic contextual information such as location and time, and implements a content-based filtering (CBF) strategy to make personalized suggestions based on the user’s tourism-related user-generated content (UGC) s/he diffuses on social media. Specifically, the CBF strategy implemented in LOOKER is based on a multi-layer user profile, where the layers representing distinct travel-related service categories (e.g., restaurants, hotels, points of interest) are modeled via language models that are defined on the basis of the captured UGC. This allows inferring the interests and the opinions of travelers about the available items. To evaluate the usefulness and the usability of the LOOKER mobile application, user studies have been conducted. The positive outcomes that have been obtained illustrate the potentials of LOOKER.

Nel dominio del turismo mobile, turisti nella ricerca di servizi correlati ai viaggi in base alla loro specifica situazione contestuale. In questa tesi, in particolare, ci concentriamo sulle due dimensioni principali del contesto attuale dell'utente: (i) il contesto spazio-temporale e (ii) il contesto cognitivo, cioè rappresentato dall'attualità dell'utente interessi.Con il miglioramento delle tecnologie context-aware, una grande quantità di fattori contestuali (tempo, luogo, ora, umore o compagno) possono essere raccolti automaticamente. Tuttavia, non tutti sono ugualmente importanti per fornire raccomandazioni efficaci. quindi, è imperativo identificare e raccogliere solo quei fattori che influenzano veramente le preferenze dell'utente e migliorare l'efficacia del sistema. Per questo motivo, come primo obiettivo di questa tesi, abbiamo proposto metodo efficace per la raccolta e la modellazione. Il nostro metodo, basato in una procedura di acquisizione del contesto selettivo, considera un fattore contestuale come pertinente se migliora la precisione del sistema. Abbiamo ampiamente esplorato il metodo proposto nel contesto del Web mobile ricerca. Un altro problema che è stato considerato in questa tesi approcci per la consulenza personalizzata sul Web (registri di ricerca, dati click-through) per definire gli interessi dell'utente. Questo tipo di le informazioni possono essere influenzate dal rumore, mentre fonti attendibili modellano gli interessi attuali dell'utente sono assolutamente necessari per avere raccomandazioni precise. Per questo motivo, in questa tesi noi focalizzato sull'estrazione degli interessi attuali dell'utente dai contenuti generati dagli utenti on-line. Più specificamente, abbiamo proposto un profilo utente multistrato, in cui ogni livello rappresenta il preferenze dell'utente in relazione a una distinta categoria di servizi correlati ai viaggi. Abbiamo usato il linguaggio statistico. Questo modello ci consente di descrivere la distribuzi, In the mobile tourism domain, travel recommendation is a crucial task that aims to assist tourists in finding relevant travel-related services based on their specific contextual situation. In this thesis, in particular, we focus on two main dimensions of the user’s current context: (i) the spatio-temporal context, and (ii) the cognitive context, i.e., represented by the user’s topical interests. With the improvement of context-aware technologies, a big amount of contextual factors can be automatically gathered. However, not all of them are equally important for providing effective recommendations. Hence, it is imperative to identify and collect only those factors that truly affect the user’s preferences and improve the system effectiveness. For this reason, as a first aim of this thesis, we proposed an effective method for gathering and modeling relevant contextual factors. Our method, based on a selective context acquisition procedure, deems a contextual factor as relevant if it improves the system accuracy. We extensively explored the proposed method in the context of mobile Web search. Another issue that has been considered in this thesis concerns the fact that traditional approaches for personalized recommendation make use of a variety of user’s historical behaviors on the Web to define the user’s interests. This kind of information may be affected by noise, while reliable sources to model the user’s topical interests are absolutely necessary to have accurate recommendations. For this reason, in this thesis we focused on the extraction of the user’s topical interests from on-line User-Generated Content. More specifically, we proposed a multi-layer user profile, where each layer represents the user’s preferences with respect to a distinct travel-related service category. We used statistical language models to model the different layers. This model enables us to depict the probability distribution of words within a user’s language that s/he employs over social med

In a ubiquitous scenario, people are typically confronted with context evolution and changing inuences. This may create new needs and may condition the user perception of what is relevant information. Over the years, different approaches have been proposed to design personalized Recommender Systems (RS), but state-of-the-art approaches mostly assume a fixed representation of a user profile; the dynamicity of the user's interests (and the way of expressing them) while interacting with the environment is not considered. Aim of this work is to predict a user's preferences in the tourism domain, to provide personalized and contextaware recommendations. Therefore, we define a user profile model which expresses in a formal way the user's opinions with respect to a particular entity. In particular, the proposed approach formally models the user generated content (UGC) connected to a group of reviews (written by expert users) for each entity, and compares it with a (positive and negative) statistical language model representing the target user profile associated with that entity. The effectiveness of the approach is illustrated on a real-case scenario.

Within Mobile information retrieval research, context information provides an important basis for identifying and understanding user’s information needs. Therefore search process can take advantage of contextual information to enhance the query and adapt search results to user’s current context. However, the challenge is how to define the best contextual information to be integrated in search process. In this paper, our intention is to build a model that can identify which contextual dimensions strongly influence the outcome of the retrieval process and should therefore be in the user’s focus. In order to achieve these objectives, we create a new query language model based on user’s pereferences. We extend this model in order to define a relevance measure for each contextual dimension, which allow to automatically classify each dimension. This latter is used to compute the degree of change in result lists for the same query enhanced by different dimensions. Our experiments show that our measure can analyze the real user’s context of up to 12000 of dimensions (related to 4000 queries). We also show experimentally the quality of the set of contextual dimensions proposed, and the interest of the measure to understand mobile user’s needs and to enhance his query.

Within Mobile information retrieval research, context information provides an important basis for identifying and understanding user’s information needs. Therefore search process can take advantage of contextual information to enhance the query and adapt search results to user’s current context. However, the challenge is how to define the best contextual information to be integrated in search process. In this paper, our intention is to build a model that can identify which contextual dimensions strongly influence the outcome of the retrieval process and should therefore be in the user’s focus. In order to achieve these objectives, we create a new query language model based on user’s pereferences. We extend this model in order to define a relevance measure for each contextual dimension, which allow to automatically classify each dimension. This latter is used to compute the degree of change in result lists for the same query enhanced by different dimensions. Our experiments show that our measure can analyze the real user’s context of up to 12000 of dimensions (related to 4000 queries). We also show experimentally the quality of the set of contextual dimensions proposed, and the interest of the measure to understand mobile user’s needs and to enhance his query.

Mobile search is a significant task in information retrieval and when coupled with context awareness technologies they can become key tools for mobile users for Web search applications. Context awareness techniques can increase the usability of mobile search providing personalized and more focussed content. However, Contextualized Mobile Information Retrieval still remains a challenging problem. This problem is to identify contextual dimensions that improve search effectiveness and should therefore be in the user's focus. We propose a context filtering process based on a new Preference Language Model, and a new relevance measurement. The experiments have been performed with over than 6000 contextual dimensions. The results show the potential of our Preference model in limiting the negative effects of contextual information overload by using the relevance measurement. © 2014 Springer International Publishing.