Graph-like data appears in many applications, such as social networks, internet hyperlinks, roadmaps, etc. and in most cases, graphs are dynamic, evolving through time. In this work, we study the problem of efficient shortest-path query evaluation on evolving social graphs. Our shortest-path queries are "temporal": they can refer to any time-point or time-interval in the graph's evolution, and corresponding valid answers should be returned. To efficiently support this type of temporal query, we extend the traditional Dijkstra's algorithm to compute shortest-path distance(s) for a time-point or a time-interval. To speed up query processing, we explore preprocessing index techniques such as Contraction Hierarchies (CH). Moreover, we examine how to maintain the evolving graph along with the index by utilizing temporal partition strategies. Experimental evaluations on real world datasets and large synthetic datasets demonstrate the feasibility and scalability of our proposed efficient techniques and optimizations.
Abstract. As the web evolves over time, the amount of versioned text collections increases rapidly. Most web search engines will answer a query by ranking all known documents at the (current) time the query is posed. There are applications however (for example customer behavior analysis, crime investigation, etc.) that would need to efficiently query these sources as of some past time, that is, retrieve the results as if the user was posing the query in a past time instant, thus accessing data known as of that time. Ranking and searching over versioned documents considers not only keyword constraints but also the time dimension, most commonly, a time point or time range of interest. In this paper, we deal with top-k query evaluations with both keyword and temporal constraints over versioned textual documents. In addition to considering previous solutions, we propose novel data organization and indexing solutions: the first one partitions data along ranking positions, while the other maintains the full ranking order through the use of a multiversion ordered list. We present an experimental comparison for both time point and time interval constraints. For time-interval constraints, different querying definitions, such as aggregation functions and consistent top-k queries are evaluated. Experimental evaluations on large real world datasets demonstrate the advantages of the newly proposed data organization and indexing approaches. If a text collection does not retain past documents, then a search query ranks only the documents as of the most current time. If the collection contains versioned documents, a search typically considers each version of a document as a separate document and the ranking is taken over all documents independently to the document's version (creation time). There are applications however, where this approach is not Introduction
Pinterest as a popular image search platform has been widely adopted by users. Every day, people come to Pinterest searching for fashion-and home decor-related content. In these domains, users exhibit stable personal tastes. In this paper, we propose a novel search algorithm which can infer user tastes from their past engagement history and tailor the search results to fit their preferences. The online and offline experiments show that our method can efficiently improve user experience and increase user engagements.
Abstract. Transaction-time databases have been proposed for storing and querying the history of a database. While past work concentrated on managing the data evolution assuming a static schema, recent research has considered data changes under a linearly evolving schema. An ordered sequence of schema versions is maintained and the database can restore/query its data under the appropriate past schema. There are however many applications leading to a branched schema evolution where data can evolve in parallel, under different concurrent schemas. In this work, we consider the issues involved in managing the history of a database that follows a branched schema evolution. To maintain easy access to any past schema, we use an XML-based approach with an optimized sharing strategy. As for accessing the data, we explore branched temporal indexing techniques and present efficient algorithms for evaluating two important queries made possible by our novel branching environment: the vertical historical query and the horizontal historical query. Moreover, we show that our methods can support branched schema evolution which allows version merging. Experimental evaluations show the efficiency of our storing, indexing, and query processing methodologies. IntroductionDue to the collaborative nature of web applications, information systems experience evolution not only on their data content but also under different schema versions. For example, Wikipedia has experienced more than 170 schema changes in its 4.5 years of lifetime [5]. Schema evolution has been addressed for traditional (single-state) database systems and issues on how data is efficiently transferred to the latest schema have been examined [4]. Consider however the case where the application maintains its past data (typically for archiving, auditing reasons etc.) which may have followed different schemas. A temporal database can be facilitated to manage the historical data, but issues related to how data can be queried under different schemas arise. The pioneering work in PRIMA system [8] addresses the issues of maintaining a transaction-time database under schema evolution by introducing: (i) an XML-based model for archiving historical data with evolving schemas, (ii) a language of atomic schema modification operators (SMOs), and (iii) query answering and rewriting algorithms for complex temporal queries spanning over multiple schema versions. Nevertheless,
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.