Prantik Bhattacharyya

Analysis of user keyword similarity in online social networks

How do two people become friends? What role does homophily play in bringing two people closer to help them forge friendship? Is the similarity between two friends different from the similarity between any two people? How does the similarity between a friend of a friend compare to similarity between direct friends? In this work, our goal is to answer these questions. We study the relationship between semantic similarity of user profile entries and the social network topology. A user profile in an on-line social network is characterized by its profile entries. The entries are termed as user keywords. We develop a model to relate keywords based on their semantic relationship and define similarity functions to quantify the similarity between a pair of users. First, we present a ‘forest model’ to categorize keywords across multiple categorization trees and define the notion of distance between keywords. Second, we use the keyword distance to define similarity functions between a pair of users. Third, we analyze a set of Facebook data according to the model to determine the effect of homophily in on-line social networks. Based on our evaluations, we conclude that direct friends are more similar than any other user pair. However, the more striking observation is that except for direct friends, similarities between users are approximately equal, irrespective of the topological distance between them.

SocialWiki: bring order to wiki systems with social context

A huge amount of administrative effort is required for large wiki systems to produce and maintain high quality pages with existing naive access control policies. This paper introduces SocialWiki, a prototype wiki system which leverages the power of social networks to automatically manage reputation and trust for wiki users based on the content they contribute and the ratings they receive. SocialWiki also utilizes interests to facilitate collaborative editing. Although a wiki page is visible to everyone, it can only be edited by a group of users who share similar interests and have a certain level of trust with each other. The editing privilege is circulated among these users to prevent/reduce vandalisms and spams, and to encourage user participation by adding social context to the revision process of a wiki page. By presenting the design and implementation of this proof-of-concept system, we show that social context can be used to build an efficient, self-adaptive and robust collaborative editing system.

Davis Social Links: Leveraging Social Networks for Future Internet Communication

In this paper, we present a social network based network communication architecture, Davis Social Links (DSL). DSL uses the trust and relationships inherent to human social networks to provide an enhanced communication architecture for future Internet designs. We begin with a conceptual discussion of how future network architectures can leverage social networks. Next, we describe the DSL architecture and how it provides to end-users control over the irreachability within the network. We conclude with a discussion of the capability to manage dynamic communities within DSL.

SIN: A Platform to Make Interactions in Social Networks Accessible

Online Social Networks (OSNs) are popular platforms for interaction, communication and collaboration between friends. In this paper we develop and present a new platform to make interactions in OSNs accessible. Most of todays social networks, including Facebook, Twitter, and Google+ provide support for third party applications to use their social network graph and content. Such applications are strongly dependent on the set of software tools and libraries provided by the OSNs for their own development and growth. For example, third party companies like CNN provide recommendation materials based on user interactions and users relationship graph. One of the limitations with this graph (or APIs) is the segregation from the shared content. We believe, and present in this paper, that the content shared and the actions taken on the content, creates a Social Interaction Network (SIN). As such, we extend Facebooks current API in order to allow applications to retrieve a weighted graph instead of Facebooks unweighted graph. Finally, we evaluate the proposed platform based on completeness and speed of the crawled results from selected community pages. We also give a few example uses of our API on how it can be used by third party applications.

Design and Implementation of Davis Social Links OSN Kernel

Social network popularity continues to rise as they broaden out to more users. Hidden away within these social networks is a valuable set of data that outlines everyones relationships. Networks have created APIs such as the Facebook Development Platform and OpenSocial that allow developers to create applications that can leverage user information. However, at the current stage, the social network support for these new applications is fairly limited in its functionality. Most, if not all, of the existing internet applications such as email, BitTorrent, and Skype cannot benefit from the valuable social network among their own users. In this paper, we present an architecture that couples two different communication layers together: the end2end communication layer and the social context layer, under the Davis Social Links (DSL) project. Our proposed architecture attempts to preserve the original application semantics (i.e., we can use Thunderbird or Outlook, unmodified, to read our SMTP emails) and provides the communicating parties (email sender and receivers) a social context for control and management. For instance, the receiver can set trust policy rules based on the social context between the pair, to determine how a particular email in question should be prioritized for delivery to the SMTP layer. Furthermore, as our architecture includes two coupling layers, it is then possible, as an option, to shift some of the services from the original applications into the social context layer. In the context of email, for example, our architecture allows users to choose operations, such as reply, reply-all, and forward, to be realized in either the application layer or the social network layer. And, the realization of these operations under the social network layer offers powerful features unavailable in the original applications. To validate our coupling architecture, we have implemented a DSL kernel prototype as a Facebook application called CyrusDSL (currently about 40 local users) and a simple communication application combined into the DSL kernel but is unaware of Facebooks API.

Mining half a billion topical experts across multiple social networks

AbstractMining topical experts on social media is a problem that has gained significant attention due to its wide-ranging applications. Here we present the first study that combines data from four major social networks—Twitter, Facebook, Google+ and LinkedIn—along with the Wikipedia graph and Internet webpage text and metadata, to rank topical experts across the global population of users. We perform an in-depth analysis of 37 features derived from various data sources such as message text, user lists, webpages, social graphs and Wikipedia. This large-scale study includes more than 12 billion messages over a 90-day sliding window and 58 billion social graph edges. Comparison reveals that features derived from Twitter Lists, Wikipedia, Internet webpages and Twitter Followers are especially good indicators of expertise. We train an expertise ranking model using these features on a large ground-truth dataset containing almost 90,000 labels. This model is applied within a production system that ranks over 650 million experts in more than 9000 topical domains on a daily basis. We provide results and examples on the effectiveness of our expert ranking system, along with empirical validation. Finally, we make the topical expertise data available through open REST APIs for wider use.

Global Entity Ranking Across Multiple Languages

We present work on building a global long-tailed ranking of entities across multiple languages using Wikipedia and Freebase knowledge bases. We identify multiple features and build a model to rank entities using a ground-truth dataset of more than 10 thousand labels. The final system ranks 27 million entities with 75% precision and 48% F1 score. We provide performance evaluation and empirical evidence of the quality of ranking across languages, and open the final ranked lists for future research.

An ‘Algorithmic Links with Probabilities’ Concordance for Trademarks: For Disaggregated Analysis of Trademark & Economic Data

Trademarks (TMs) shape the competitive landscape of markets for goods and services in all countries through branding and conveying information and quality inherent in products. Yet, researchers are largely unable to conduct rigorous empirical analysis of TMs in the modern economy because TM data and economic activity data are organized differently and cannot be analyzed jointly at the industry or sectoral level. We propose an ‘Algorithmic Links with Probabilities’ (ALP) approach to match TM data to economic data and enable these data to speak to each other. Specifically, we construct a NICE Class Level concordance that maps TM data into trade and industry categories forward and backward. This concordance allows researchers to analyze differences in TM usage across both economic and TM sectors. In this paper, we apply this ALP concordance for TMs to characterize patterns in TM applications across countries, industries, income levels and more. We also use the concordance to investigate some of the key determinants of international technology transfer by comparing bilateral TM applications and bilateral patent applications. We conclude with a discussion of possible extensions of this work, including deeper indicator-level concordances and further analyses that are possible once TM data are linked with economic activity data.

Information Flow and Search in Unstructured Keyword Based Social Networks

In online social networks (OSNs), user connections can be represented as a network. The network formed has distinct properties that distinguish it from other network topologies. In this work, we consider an unstructured keyword based social network topology where each edge has a trust value associated with it to represent the mutual relationship between the corresponding nodes. Users have keywords as their profile attributes that have policies associated with them to define abstractly the flow of keyword information and the accessibility to other users in the network. We also address privacy concerns as outlined in works on future OSN architectures.This paper makes two key contributions. First, we develop an information flow model to disseminate keyword information when users add keywords as their profile attributes. Second, for keyword based queries, we design and develop a search algorithm to find users with the specified keywords in their profile attributes. It is based on a linear combination of topological distance and trust metrics. It is also dynamic in nature such that it adapts itself for each individual node during the search process. We observe an improvement in orders of magnitude when the search algorithm is compared to breadth first search.

An ‘Algorithmic Links with Probabilities’ Concordance for Trademarks with an Application Towards Bilateral IP Flows

Trademarks (TMs) shape the competitive landscape of markets for goods and services in all countries. As a key element of branding, they can inform consumers about the quality and content of goods and services. Yet, researchers are largely unable to conduct rigorous empirical analysis of TMs in the global economy because TM data and economic data are organised differently and cannot be analysed jointly at the industry or sector level. We propose an ‘algorithmic links with probabilities’ (ALP) approach to match TM data to economic data and enable joint analysis with these data. Specifically, we construct a NICE class-level concordance that maps TM data into trade and industry categories forward and backward. This concordance allows researchers to analyse differences in TM usage across both economic and TM sectors. We apply this ALP concordance for TMs to characterise patterns in TM registrations across countries and industries and to investigate some key determinants of international technology flows by comparing bilateral TM registrations and bilateral patent grants. We find that international patenting and TM flows are jointly determined by trade-related influences with significant differences in intellectual property usage across industry sectors and income levels.