In this big data era, knowledge becomes increasingly linked, along with the rapid growth in data volume. Connected knowledge is naturally represented and stored as knowledge graphs, which are of more and more importance for many frontier research areas such as machine intelligence. Effectively finding relations between entities in a large knowledge graph plays a key role in many knowledge graph applications, as the most valuable part of a knowledge graph is its rich connectedness, which captures rich information about the objects in the real world. However, due to the intrinsic complexity of real-world knowledge, finding semantically close relations by navigation in a large knowledge graph is very challenging. Canonical graph exploration methods inevitably result in combinatorial explosion especially when the paths connecting two entities are long: the search space is$O(d^l)$, where$d$is the average graph node degree and$l$is the length of the path. In this paper, we will systematically study the semantic navigation problem for large knowledge graphs. Inspired by AlphaGo, which was overwhelmingly successful in the game Go, we designed an efficient semantic navigation method based on a well-tailored Monte Carlo Tree Search algorithm with the unique characteristics of knowledge graphs considered. Extensive experiments on different real-life knowledge bases show that our method is not only effective but also very efficient.

The address space transformation (AST) algorithm is proposed to solve the classical consistency maintenance problem in real-time collaborative editing systems, which is widely adopted in web 2.0, mobile comment and itinerary planning applications. None of the state-of-art AST approaches, however, support string-wise operations natively. They normally split a string into individual characters, thus eliciting low efficiency when user-based operations such as copy and paste are executed. In this paper, we propose a string-wise address space transformation algorithm called ASTS which mainly aims to address two challenges as follows: (1) How to construct a simple model for the string-wise operation. (2) How to maintain stable and high response algorithm performance. Two experiments are given to verify the adopted algorithm's effectiveness in diverse perspective. Compared with traditional AST algorithm, our experiments show that ASTS has improved in string-wise operation environment and also performs well in character-wise environment.

RDF is one of the most commonly used knowledge representation forms. Many highly influential knowledge bases, such as Freebase and PubChemRDF, are in RDF format. An RDF data set is usually represented as a collection of subject-predicate-object triples. Despite the flexibility of RDF triples, it is challenging to serve SPARQL queries on RDF data efficiently by directly managing triples due to the following two reasons. First, heavy joins on a large number of triples are needed for query processing, resulting in a large number of data scans and large redundant intermediate results; Second, weakly-typed triple representation provides suboptimal random access - typically with logarithmic complexity. This data access challenge, unfortunately, cannot be easily met by a better query optimizer as large graph processing is extremely I/O-intensive. In this paper, we argue that strongly-typed graph representation is the key to high-performance RDF query processing. We propose Stylus - a strongly-typed store for serving massive RDF data. Stylus exploits a strongly-typed storage scheme to boost the performance of RDF query processing. The storage scheme is essentially a materialized join view on entities, it thus can eliminate a large number of unnecessary joins on triples. Moreover, it is equipped with a compact representation for intermediate results and an efficient graph-decomposition based query planner. Experimental results on both synthetic and real-life RDF data sets confirm that the proposed approach can dramatically boost the performance of SPARQL query processing.

With the development of information technology, a growing number of users begin to use Internet as a platform to plan their itinerary collaboratively. During the online itinerary planning process, collaboration plays an increasingly significant role. As the number of users grows, it is challenging to support more and more participants to plan itineraries online collaboratively, due to the requirement of responsiveness and data consistency. Traditional concurrency control methods (e.g. lock-based protocols, etc.) are costly because their primary copies are on a dedicated server and network latency, and concurrent operations violate CCI model. Address Space Transformation (AST) proves to the high performance in addressing the issue of interactive responsiveness and consistency maintenance. In this paper, a novel AST based approach is presented to support collaborative itinerary planning. We report how we address the emerging consistency maintenance challenges in directed graph of online itinerary planning. With this method, users are allowed to participate and leave itinerary planning tasks anytime. Moreover, collaborative itinerary planning also can be modified in real time with data consistent guarantees. A system called CLIP is implemented to evaluate its performance. Experimental results demonstrate the benefit and performance of this approach over the existing methods.

Commenting systems play increasingly important roles in the interactive web applications. Meanwhile, more and more web applications are visited on mobile devices. However, the intermittent connection of mobile networks and resource limitation of mobile devices pose great challenges, mainly in terms of interactive responsiveness and data consistency. In this paper, we present the first work of partial replication solution based on collaborative editing techniques, which can address the issues of local responsiveness and resource limitation on mobile commenting systems. We report how we address the consistency maintenance challenges that come with the partial replication approach. With this approach, users are allowed to smoothly comment anywhere anytime. The comment thread can be incrementally updated and automatically synchronized with strong data consistency guarantees. We implemented a system prototype called Hydra and evaluated it on a real data set.

With the prevalence of cloud services, more and more editing tools are moving from client to cloud for collaboration. Google Docs, Microsoft Office Web Apps, Zoho Office Suite are such representative cloud backed co-editing services. People from different organizations may choose different service providers, which implies the needs of cross-cloud document collaboration. However, it is hard to make different collaborative editing services interoperate with each other. On the one hand, it is not realistic to let all service providers comply to a unified standard and provide interoperation interfaces. On the other hand, the challenge of consistency maintenance in real-time document synchronization still exists. These make interoperability one of the most significant collaboration barriers in online collaborative editing. This paper presents a transparent approach called Hermes to interoperating between heterogeneous collaborative editing services in the cloud. Users are allowed to use their familiar services to participate in the cross-cloud document collaboration. Service providers do not need to put any effort to interoperate with other services. Our approach is validated by a system prototype.

Content consuming and sharing are two most important user activities in social networking sites (SNSs). Lots of studies have been conducted on content recommendation using users' common interests. However, little has been done to help users to select friends and share content within their social networks. In this paper, we contribute a recommendation framework AOPUT to recommend both content and friend list for sharing to users leveraging content and social information in SNSs. It consists of two recommendation components: Recder and ShareAider. Recder generates content recommendations by connecting users with common interests. An improved Jaccard similarity is proposed to improve the Collaborative Filtering (CF) recommendation quality. ShareAider recommends a friend list to users when they want to share content with their friends. CF method and a social-based method are compared and the combination of them are explored to achieve better results. AOPUT is evaluated on a real world social network. The experimental results show that (1) Recder can provide better recommendation quality than the traditional CF method thanks to the improved Jaccard similarity; (2) social-based method performs better than CF since the sharing behavior in SNSs are highly dominated by users' social preferences, and the combination of these two methods performs better than each of them individually.

This paper discusses issues of infrastructure and user experiences based on an empirical study of the online gaming platform ArcOnline in China. ArcOnline has allowed millions of users to play and watch arcade games over the Internet. What the study of ArcOnline helps bring to the fore are the infrastructural aspects of user experiences, mainly because the high demand for responsiveness and resources of arcade gaming has made infrastructures visible. In this paper, we present our findings of how the experience of playing arcade games changes with different socio-technical infrastructures, the arcade room and the online environment, highlighting how the form of the infrastructure shapes experience and collaboration, as well as how new social infrastructures are emerging through the interplay between the game, the platform, the community and the media ecology. We end by discussing the importance of considering infrastructure in understanding and designing new media experiences.

With the explosive growth of online social communities and massive user-generated content, privacy-preserving recommender systems, which identify information of interest to individual users without disclosing personal interests to other parties, have become increasingly important. Collaborative filtering (CF), a widely used recommendation technique, recommends content that similar users have liked. As a result, CF-based recommender systems may expose sensitive personal interest information. This is demonstrated by a privacy attack model we present that targets online social communities. To solve this problem, we propose an interest group based privacy-preserving recommender system called Pistis. By identifying inherent item-user interest groups and separating users' private interests from their public interests, Pistis can make recommendations based on aggregated judgments of group members and local personalization, thus avoiding the disclosure of personal interest information. Pistis has been deployed and evaluated in an online social community with over 63,000 users, 20,000 daily posts, and 180,000 daily reads. Compared with two representative CF-based methods, our evaluation results demonstrate that Pistis achieves better performance in privacy preservation, recommendation quality, and efficiency.