Rabeeh Ayaz Abbasi

Exploiting Flickr Tags and Groups for Finding Landmark Photos

Many people take pictures of different city landmarks and post them to photo-sharing systems like Flickr. They also add tags and place photos in Flickr groups, created around particular themes. Using tags, other people can search for representative landmark images of places of interest. Searching for landmarks using tags results into many non-landmark photos and provides poor landmark summary for a city. In this paper we propose a new method to identify landmark photos using tags and social Flickr groups. In contrast to similar modern systems, our approach is also applicable when GPS-coordinates for photos are not available. Presented user study shows that the proposed method outperforms state-of-the-art systems for landmark finding.

RichVSM: enRiched vector space models for folksonomies

People share millions of resources (photos, bookmarks, videos, etc.) in Folksonomies (like Flickr, Delicious, Youtube, etc.). To access and share resources, they add keywords called tags to the resources. As the tags are freely chosen keywords, it might not be possible for users to tag their resources with all the relevant tags. As a result, many resources lack sufficient number of relevant tags. The lack of relevant tags results into sparseness of data, and this sparseness of data makes many relevant resources unsearchable against user queries. In this paper, we explore two dimensions of semantic relationships between tags, based on the context and the distribution of tags. We exploit semantic relationships between tags to reduce sparseness in Folksonomies and propose different enriched vector space models. We also propose a vector space model Best of Breed which utilizes appropriate enrichment method based on the type of the query. We evaluate the proposed methods on a large dataset of 27 million resources, 92 thousand tags and 94 million tag assignments. Experimental results show that the enriched vector space models help in improving search, especially for the rare queries which have few relevant resources in the sparse data.

Saving lives using social media: Analysis of the role of twitter for personal blood donation requests and dissemination

Abstract Social media has an impact on many aspects of human life ranging from sharing personal information to revolutionizing political systems of entire countries. One not so well studied aspect of social media is analyzing its usage and efficacy in healthcare, particularly in developing countries which lack state-of-the-art healthcare systems and processes. In such countries, social media may be used to facilitate patient-centric healthcare by involving the patient for fulfilling personal healthcare needs. This article provides an in-depth analysis of one such need, that is, how people use social media to request for blood donations. We study the request and dissemination behavior of people using social media to fulfill blood donation requests. We focus on twitter, and blood donation accounts in India. Our study reveals that each of the seven twitter accounts we studied have a large followership of more than 35,000 users on an average and receive a substantial number (more than 900) of donation requests in a day on an average. We analyze the requests in various ways to present an outlook for healthcare providers to make their systems more patient-centric through a better understanding of the needs of people requesting for blood donations. Our study also identifies areas where future social media enabled automated healthcare systems can focus on the needs of individual patients. These systems can provide support for saving more lives by reducing the gap between blood donors and the people in need.

Large Scale Tag Recommendation Using Different Image Representations

Nowadays, geographical coordinates (geo-tags), social annotations (tags), and low-level features are available in large image datasets. In our paper, we exploit these three kinds of image descriptions to suggest possible annotations for new images uploaded to a social tagging system. In order to compare the benefits each of these description types brings to a tag recommender system on its own, we investigate them independently of each other. First, the existing data collection is clustered separately for the geographical coordinates, tags, and low-level features. Additionally, random clustering is performed in order to provide a baseline for experimental results. Once a new image has been uploaded to the system, it is assigned to one of the clusters using either its geographical or low-level representation. Finally, the most representative tags for the resulting cluster are suggested to the user for annotation of the new image. Large-scale experiments performed for more than 400,000 images compare the different image representation techniques in terms of precision and recall in tag recommendation.

CC-GA: A clustering coefficient based genetic algorithm for detecting communities in social networks

Abstract A community structure is an integral part of a social network. Detecting such communities plays an important role in a wide range of applications, including but not limited to cluster analysis, recommendation systems and understanding the behaviour of complex systems. Researchers have derived many algorithms to discover the community structures of networks. Discovering communities is a challenging task, and there is no single algorithm that produces the best results for all networks. Therefore, despite many elegant solutions, discovering communities remains an active area of research. In this paper, we propose a novel algorithm, the Clustering Coefficient-based Genetic Algorithm (CC-GA), for detecting them in social and complex networks. Researchers have used several genetic algorithms to detect communities, but the proposed algorithm is novel in terms of both the generation of the initial population and the mutation method, and these improve its efficiency and accuracy. Experiments on a variety of real-world datasets and a comparison to state-of-the-art genetic and non-genetic-based algorithms show improved results.

Query expansion in folksonomies

People share resources in folksonomies and add tags to these resources. There are often only a few tags associated with each resource, which makes the data available in folksonomies extremely sparse. Spareness in folksonomies makes searching resources difficult. Many relevant resources against a query might not be retrieved if they are not associated with the queried terms. One possible way to overcome the search problem in folksonomies is query expansion. We propose and compare different query expansion techniques for folksonomies and evaluate these methods on a large scale. We also propose an automated evaluation method for query expansion. Experimental results show that query expansion helps in improving search in folksonomies.

CommuniMents: A Framework for Detecting Community Based Sentiments for Events

Social media has revolutionized human communication and styles of interaction. Due to its effectiveness and ease, people have started using it increasingly to share and exchange information, carry out discussions on various events, and express their opinions. Various communities may have diverse sentiments about events and it is an interesting research problem to understand the sentiments of a particular community for a specific event. In this article, the authors propose a framework CommuniMents which enables us to identify the members of a community and measure the sentiments of the community for a particular event. CommuniMents uses automated snowball sampling to identify the members of a community, then fetches their published contents (specifically tweets), pre-processes the contents and measures the sentiments of the community. The authors perform qualitative and quantitative evaluation for a variety of real world events to validate the effectiveness of the proposed framework.

Prototype selection to improve monotonic nearest neighbor

Abstract Student surveys occupy a central place in the evaluation of courses at teaching institutions. At the end of each course, students are requested to evaluate various aspects such as activities, methodology, coordination or resources used. In addition, a final qualification is given to summarize the quality of the course. The prediction of this final qualification can be accomplished by using monotonic classification techniques. The outcome offered by these surveys is particularly significant for faculty and teaching staff associated with the course. The monotonic nearest neighbor classifier is one of the most relevant algorithms in monotonic classification. However, it does suffer from two drawbacks, (a) inefficient execution time in classification and (b) sensitivity to no monotonic examples. Prototype selection is a data reduction process for classification based on nearest neighbor that can be used to alleviate these problems. This paper proposes a prototype selection algorithm called Monotonic Iterative Prototype Selection (MONIPS) algorithm. Our objective is two-fold. The first one is to introduce MONIPS as a method for obtaining monotonic solutions. MONIPS has proved to be competitive with classical prototype selection solutions adapted to monotonic domain. Besides, to further demonstrate the good performance of MONIPS in the context of a student survey about taught courses.

INTWEEMS: a framework for incremental clustering of tweet streams

Twitter is a popular micro-blogging service for sharing short messages called tweets. Tweets provide public opinion on various topics. Currently twitter presents search results in form of a flat list, sorted either by popularity or by recency. These search results limit the possibility of identifying diverse latent topics covered by the tweets. One way to better understand the tweets is to cluster them where each cluster depicts a latent topic. Suitable clustering algorithms are required to cluster streaming data and map new data into existing clusters. To address this, we propose in this paper a framework called INTWEEMS (INcremental clustering of TWEEt streaMS) which clusters tweets in real-time, adjusts new tweets into existing clusters (incrementally), and provides visualization of clusters that helps in identifying latent topics and sub-topics within the tweets. This paper describes the INTWEEMS framework and its implementation.

Who will cite you back? Reciprocal link prediction in citation networks

Purpose Link prediction in social networks refers toward inferring the new interactions among the users in near future. Citation networks are constructed based on citing each other papers. Reciprocal link prediction in citations networks refers toward inferring about getting a citation from an author, whose work is already cited by you. The paper aims to discuss these issues. Design/methodology/approach In this paper, the authors study the extent to which the information of a two-way citation relationship (called reciprocal) is predictable. The authors propose seven different features based on papers, their authors and citations of each paper to predict reciprocal links. Findings Extensive experiments are performed on CiteSeer data set by using three classification algorithms (decision trees, Naive Bayes, and support vector machines) to analyze the impact of individual, category wise and combination of features. The results reveal that it is likely to precisely predict 96 percent of reciprocal links. The study delivers convincing evidence of presence of the underlying equilibrium amongst reciprocal links. Research limitations/implications It is not a generic method for link prediction which can work for different networks with relevant features and parameters. Practical implications This paper predicts the reciprocal links to show who is citing your work to collaborate with them in future. Social implications The proposed method will be helpful in finding collaborators and developing academic links. Originality/value The proposed method uses reciprocal link prediction for bibliographic networks in a novel way.