Brazil

The current advances have recognized the online recommendation task as a Multi-Armed Bandit (MAB) problem and proposed several algorithms by combining it with Collaborative Filtering (CF). In general, their effectiveness is related to their practical ability to handle the trade-off between selecting the most relevant item (exploitation) or choose an unexpected item to add more knowledge to the system (exploration). However, we observed that a naive failure has been introduced in the main CF bandit models, which have intrinsically reduced itself to pure exploitation of the information known a prior by the system (e.g., items’ popularity) or a pure exploration of the items in the first user’s interaction. Conversely, we hypothesize this trade-off should be considered since the first user’s interactions to improve the system’s effectiveness. Thus, we propose a new methodology to measure the impact of this problem. This methodology has two main phases. In phase 1, traditional non-personalized methods resemble pure-exploitation and pure-exploration to feed the learning of all MAB models. Then, analyzing the MAB effectiveness in phase 2, we may determine the best approach to be considered in the first user’s interactions. Moreover, we also propose and add a new strategy that combines exploration and exploitation in phase 1. Indeed, our experimental results show that the combination of exp-exp at first user’s interactions improves the system’s effectiveness without adding more effort to users.

WebMedia 2020

The impact of first recommendations based on exploration or exploitation approaches in recommender systems’ learning

cold-start problem

multi-armed bandits

collaborative filtering

ICPR 2020 is the premier international conference on Pattern Recognition, gathering leading researchers from around the world to present their latest achievements in Pattern Analysis and Recognition, applied Artificial Intelligence and Machine Learning. The topic of the year is " Putting Artificial intelligence to work on Patterns". 
For the first time, ICPR has employed a two-round review process similar to journal submissions, with new papers allowed to be submitted in either the first or second round and papers submitted in the first round and not accepted allowed to be revised and re-submitted for second round review. 

In this edition of the **ICPR 2020** Conference, there are 1407 Oral/Poster presentations organized in five Tracks: 
- Artificial Intelligence, Machine Learning for Pattern Analysis; 
- Biometrics, Human Analysis and Behavior Understanding; Computer Vision, Robotics and Intelligent Systems; 
- Document and Media Analysis; and 
- Image and Signal Processing. 

The **ICPR 2020** Main Conference program also includes 7 Keynotes from prominent world-renowned scientists, 8 Tutorials, 28 Demos, 8 Challenges, 2 Panels, as well as an Industry/Sponsor Exhibition. In addition, 41 Workshops on a range of important current and emerging research topics complete the ICPR 2020 program. 


ICPR 2020

technical paper

Promoted by the Brazilian Computer Society (SBC), the Brazilian Symposium on Multimedia and Web (WebMedia) is the main event of the theme in Brazil and an excellent opportunity for scientific and technical exchange among students, researchers and professionals in the areas of Multimedia, Hypermedia and Web. This year we will be celebrating to 26th edition of WebMedia, which will be held online from november 30th to December 4th .

Webmedia 2020’s programming includes:

– Technical sessions for the presentation of scientific papers
– National and international keynotes
– Panel with companies
– Workshops
– Short Courses
– Open community meetings

The topics of interest for submitting papers are:

– Web and Social Networks
– Multimedia
– Digital TV, Ubiquitous and Mobile Computing

Besides the main track, WebMedia 2020 will have in parallel the Tools and Applications Workshop, works Tools and Applications Workshop, the Workshop on Undergraduate Research Work and the Workshop The Future of Interactive.

The WebMedia 2020 organization is under the responsibility of the Programa de Pós-graduação Doutorado em Ciência da Computação – Associação UFMA/UFPI (DCCMAPI), do Programa de Pós-graduação em Ciência da Computação/ CCET (PPGCC), do Laboratory of Advanced Web Systems/UFMA (LAWS) e do Laboratório de Sistemas Multimídia/UFMA (TELEMÍDIA).).

#PartiuWebMedia2020 #WebMedia2020

Promoted by the Brazilian Computer Society (SBC), the Brazilian Symposium on Multimedia and Web (WebMedia) is the main event of the theme in Brazil and an excellent opportunity for scientific and technical exchange among students, researchers and professionals in the areas of Multimedia, Hypermedia and Web.

Location-Based Social Networks (LBSNs) have become important tools for people interested in exploring new places. And, similar to traditional recommendation domains, handling the trade-off between accuracy and diversity is a major challenge to provide useful recommendations. However, this domain adds an equally relevant dimension to this challenge: the geographical distance between users and each point-of-interest (POI). Besides understanding how the characteristics of services offered by each POI fit the user needs, realizing how far users are willing to move to fulfill these needs is of paramount relevance. Moreover, the users present distinct levels of interest in diversification. In this paper, we propose a strategy to provide POI recommendations linearly combining categorical and geographical diversifications in a personalized way. Indeed, our strategy is able to prioritize quality dimensions that better suit the personalized needs of each user, with gains up to 10% when compared with unpersonalized versions.

Combining complementary diversification models for personalized POI recommendations

More than just a common habit or pastime of people, listening to music is often used as a way to facilitate listeners to achieve a certain mental state or to perform specific activities. Among the most common activities performed when listening to music is relaxation. According to findings from the area of music therapy, a good musical choice for relaxation needs personalization. However, until now, many studies in the area of Music Information Retrieval (MIR) have focused on only identifying general characteristics of music according to certain contexts. Thus, this paper focuses on expanding the findings of music therapy, considering aspects of musicology and a large amount of data. To this end, we: (i) explore more than 60 thousand handcrafted playlists of almost 5 thousand users; (ii) create a user preference model based on high and low-level features of the songs and (iii) analyze the differences between the users' preferences. The results of our analysis suggest that personalization plays a key role in the playlists created by the users. With this work, we want to pave the way and encourage the improvement of context-aware music recommender systems or other MIR concepts like user behavior analysis.

Investigating the Role of Personalization When Creating Relaxing Playlists

Currently, videos play an important role in people's every day lives and there is a continuous increase in their production and consumption. However, to make video information access a feasible task, smarter and more efficient strategies are needed in order to deal with this complex and huge data volume. In this context, automatic video summarization becomes a significant tool. This research area aims to produce a smaller version of original video, keeping its most meaningful information. Particularly for some relavant domains, like news and education, it is important a summary to be generated from a set of related videos - different videos about the same subject. In the automatic multi-video summarization field a common and essential step is to define content selection criteria. A promising approach, yet unexploited by related work, is to apply inclusion and exclusion criteria derived from human strategies. In this work, we investigate how exclusion criteria can help to keep multi-video summaries short and informative. We introduce an approach to detect subjectivity into video segments from news domain, allowing those segments to be not included into the summary. Our approach is evaluated by the means of objective measures of effectiveness. We also analyze the impact that removing such subjective segments has on multi-video summarization outputs by conducting a study with final users. Results indicate that the proposed approach is able of producing more compact summaries, keeping the relevant content for the users.

Investigating Subjectivity Criterion for Multi-video Summarization

The recent availability of HMD devices (Head-Mounted Display) spurred advances in Virtual Reality and Augmented Reality. It allowed scenarios of immersion of users in realistic environments, e.g., synthesized 3D environments for games and simulation, and omnidirectional videos, namely 360° videos. These HMD devices are commonly used in conjunction with home computers or in controlled environments (e.g., museums). However, despite these scenarios, current interactive TV systems still focus on using interactive content in two-dimensional layouts. These systems could use HMDs to provide immersive experiences to viewers. This paper, therefore, presents Guaraná, a player for interactive 360° scenes using HMDs as second screen devices. Guaraná controls a 360 scene execution based on external control from a TV set. Thus, it enables its use in conjunction with an intereactive multimedia application.

Interactive 360-degree Videos in Ginga-NCL UsingHead-Mounted-Displays as Second Screen Devices

Smart city services are typically defined according to domains (e.g., health, education, safety) and supported by different systems. Consequently, the analysis of smart city data is often domain-specific, thus limiting the capabilities of the offered services and hampering decision-making that relies on isolated domain information. To support a suitable analysis across multiple domains, it is necessary having a unified data model able to handle the inherent heterogeneity of smart city data and take into account both geographic and citizen information. This presentation is related to an ontology-based information model to support multi-domain analysis in smart cities to foster interoperability and powerful automated reasoning upon unambiguous information. The semantic relationships and properties defined in the model also allow inferring new pieces of information that improve accuracy when analyzing multiple city domains. An evaluation of the information model through ontological metrics and competence questions also will be presented.

An Ontology-based Information Model for Multi-Domain Semantic Modeling and Analysis of Smart City Data

On social media platforms, comments associated with news pieces are usually filled with negativity and toxicity, many times promoting flamed discussions and insults among users. Although designed to encourage conversations and interactions, the high toxicity might end up contributing to create a hostile environment in the online space, which is detrimental to both social media platforms and their users. In this work, we provide a diagnostic about the toxicity in comments associated with news shared on Facebook. To do that, we collected all posts and comments from relevant pages during a major political event in Brazil, the release of Former President Lula from prison. We then used the Perspective API from Google to measure the toxicity of the comments and posts. Our analysis of the toxicity unveils features that influence toxicity associated with the news, especially in relation to public figures. We hope our findings may affect the design of better content policies able to mitigate the problem.

Characterizing Toxicity on Facebook Comments in Brazil

In view of the emergence of mobility restrictions and social isolation imposed by the coronavirus or COVID-19 pandemic, digital media, especially social networks, become a breeding ground for fake news, political attacks and large-scale misinformation. The impacts of this `infodemic' can take even greater proportions when using sponsored content on social networks, such as Facebook ads. Using the Facebook ad library we collected more than 236k facebook ads from 75 different countries. Choosing ads from Brazil as the focus of research, we found ads with political attacks, requests for donations, doctors prescribing vitamin D as a weapon to fight coronavirus, among other contents with evidence of misinformation.

Analyzing the Use of Covid-19 Ads on Facebook

In the last decades, the dissemination of News through digital media has increased the information accessibility previously offered by traditional channels. Despite their benefits, digital media have exacerbated an old problem: the spread of Fake News, (i.e., false News intentionally published). Faced with this scenario, the linguistic approaches to automatic Fake News detection use information that can be directly extracted from the News' text. Several methods based on these approaches use grammatical classification and sentiment analysis over News writing in Portuguese. However, as far as it was possible to observe in the related literature, these methods are limited to the identification of polarity of sentiment (i.e., positive, neutral or negative) existing in the text. Although polarity classification be an effective method for a wide range of natural language processing applications, it does not address language nuances (e.g., emotions such as anger, sadness, etc.) that can provide evidence that a text contains false information. Hence, this study proposes an extended method that, in addition to the grammatical classification and polarity based sentiment analysis, also uses the analysis of emotions to detect Fake News written in Portuguese. The extended method showed promising results in experimental data, obtaining accuracy greater than 92\%. In average, the proposed method overcame polarity and gramatical classification based methods in 1.4 percentage points.

A Linguistic-Based Method that Combines Polarity, Emotion and Grammatical Characteristics to Detect Fake News in Portuguese

As smartphones increased in popularity, mobile devices became a crucial technological element in Information Systems. Beyond the end-user platform, smartphones have several sensors to capture and characterise the user's context. Current Context-Aware Mobile (CAM) applications use this data to improve user experience by filtering data, services, and, even, adapting the system behavior. The design and development of the capture, inference, and action parts of CAM applications can be challenging for mobile developers, due to the diversity of mobile hardware ecosystem and sensor APIs' complexity. This paper presents EasyContext, a visual tool for designing and generating contextual rules for mobile devices, using the Google Awareness API as its contextual information provider. Mobile developers specify what to capture and when to react to context-changes in the visual tool and these contextual rules are exported to Android projects. We evaluated our approach with eight developers, which were selected to participate in a quasi-experiment. The experiment involves developing two mobile applications using both EasyContext and Google Awareness API. Preliminary results show that it is possible to build applications with fewer lines of code compared to the Google Awareness API. According to the developers, the proposed approach has better readability and makes it easier to design contextual rules.

A Generative Approach for Android Sensor-based Applications

Research indicates that people can receive more useful product recommendations if the filtering process considers their personality. In this paper, we propose a hybrid strategy for Recommender Systems (using matrix factorization and personality-based neighborhood) to recommend the best products calculated for a particular customer (user). The proposed user profile used in the definition of the neighborhood involves these three personality models: Big Five (or OCEAN, or Five-Factor Model), Needs, and Values. We experimented with data from more than 10,000 Amazon customers. We inferred their personality characteristics from the analysis of reviews via IBM Watson Personality Insights. The results indicated that the proposed strategy's performance was better than that of the state-of-the-art algorithms analyzed. Besides, there was no statistical difference between using only the Big Five model or using it together with the Needs and Values models.

The impact of first recommendations based on exploration or exploitation approaches in recommender systems’ learning

Downloads

Next from WebMedia 2020

Combining complementary diversification models for personalized POI recommendations

Similar lecture

Probabilistic Latent Factor Model for Collaborative Filtering with Bayesian Inference

The impact of first recommendations based on exploration or exploitation approaches in recommender systems’ learning

.css-70qvj9{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;}Downloads

Next from WebMedia 2020

Combining complementary diversification models for personalized POI recommendations

Similar lecture

Probabilistic Latent Factor Model for Collaborative Filtering with Bayesian Inference

Downloads