We investigate the data management of temporal property graphs on atemporal Labeled Property Graph (LPG) engines. While existing frameworks define temporal semantics, practitioners lack systematic empirical guidance on mapping these models to concrete representations in the absence of native engine support. We formalize two fundamental representation strategies implemented on top of standard LPG systems: a Versioning strategy that materializes temporal states via topological expansion, and a Property Lists strategy that encapsulates evolution through property-level compression. For each strategy, we define query translations for temporal evolution operators and analytical aggregations, and evaluate them across two heterogeneous real-world datasets. Our evaluation reveals a fundamental trade-off between topological selectivity and historical compactness: Versioning excels at high-throughput ingestion and short-window workloads but suffers from traversal amplification in dense graphs over long horizons, while Property Lists amortizes scan costs for read-heavy, multi-window analytics but incurs higher ingestion overhead and cannot support time-varying labels. We identify workload-dependent crossover points, showing how dataset density and structural volatility determine the superior strategy.

When the focus is on the relationships or interactions between entities, graphs offer an intuitive model for many real-world data. Such graphs are usually large and change over time, thus, requiring models and strategies that explore their evolution. We study the evolution of aggregate graphs and introduce the GraphTempo model that allows temporal and graph aggregation not only on node level by grouping individual nodes, but on a pattern level as well, where subgraphs are grouped together. Furthermore, we propose an efficient strategy for exploring the evolution of the graph based on identifying time intervals of significant growth, shrinkage, or stability. Finally, we evaluate the efficiency and effectiveness of the proposed approach using four real graphs.

Soundex has been used for over a century for approximately matching records based on their phonetic footprint. In this paper, we examine a series of techniques a practitioner might employ in order to increase the algorithm’s matching capabilities, when utilizing Soundex for privacy preserving record linkage and a protocol based on Apache Spark, suitable for big data processing. We provide a detailed empirical assessment measuring matching quality and time performance of the proposed alternatives, showing that we achieve both precision and recall over 95% for large datasets in a few seconds and without utilizing any privacy-preserving blocking technique.

When dealing with sensitive and personal user data, the process of record linkage raises privacy issues. Thus, privacy preserving record linkage has emerged with the goal of identifying matching records across multiple data sources while preserving the privacy of the individuals they describe. The task is very resource demanding, considering the abundance of available data, which, in addition, are often dirty. Blocking techniques are deployed prior to matching to prune out unlikely to match candidate records so as to reduce processing time. However, when scaling to large datasets, such methods often result in quality loss. To this end, we propose Multi-Sampling Transitive Closure for Encrypted Fields (MS-TCEF), a novel privacy preserving blocking technique based on the use of reference sets. Our new method effectively prunes records based on redundant assignments to blocks, providing better fault-tolerance and maintaining result quality while scaling linearly with respect to the dataset size. We provide a theoretical analysis on the method's complexity and show how it outperforms state-of-the-art privacy preserving blocking techniques with respect to both recall and processing cost.

The growth of online services has created the need for duplicate elimination in high-volume streams of events. The sheer volume of data in applications such as pay-per-click clickstream processing, RSS feed syndication and notification services in social sites such Twitter and Facebook makes traditional centralized solutions hard to scale. In this paper, we propose an approach based on distributed filtering. To this end, we introduce a suite of distributed Bloom filters that exploit different ways of partitioning the event space. To address the continuous nature of event delivery, the filters are extended to support sliding window semantics. Moreover, we examine locality-related tradeoffs and propose a tree-based architecture to allow for duplicate elimination across geographic locations. We cast the design space and present experimental results that demonstrate the pros and cons of our various solutions in different settings.

In this paper, we address the problem of database selection for XML document collections, that is, given a set of collections and a user query, how to rank the collections based on their goodness to the query. Goodness is determined by the relevance of the documents in the collection to the query.

Due to the structural heterogeneity of XML, queries are often interpreted approximately. This is achieved by relaxing the query and ranking the results based on their relevance to the original query. Query relaxation over distributed XML repositories may incur large communication costs, since partial result lists from different sites need to be gathered and ranked to assembly the overall top-k results. To process such queries efficiently, we propose using a distributed clustered index to group documents based on their structural similarity. The clustered index proves to be very effective in reducing the sizes of the partial lists that need to be combined. Furthermore, it can be used as the basis of a pay-as-you-go approach, where clusters of documents are accessed gradually providing the user with increasingly improving results. To reduce the cost of constructing and maintaining the clustered index, we use a compact data structure that trades-off accuracy for storage and communication efficiency. The index is also used for selectivity estimation so that query relaxation is geared towards the most promising structural transformations. Our experimental results show that our approach significantly reduces the communication cost for retrieving the top-k results, while maintaining a low construction cost for the clustered index.

In many large-scale content sharing applications, participants or peers are grouped together forming clusters based on their content or interests. In this paper, we deal with the maintenance of such clusters in the presence of updates. We model the evolution of the system as a strategic game, where peers determine their cluster membership based on a utility function of the query recall. Peers are guided either by selfish or altruistic motives: selfish peers aim at improving the recall of their own queries, whereas altruistic peers aim at improving the recall of the queries of other peers. We study the evolution of such clusters both theoretically and experimentally under a variety of conditions. We show that, in general, local decisions made independently by each peer enable the system to adapt to changes and maintain the overall recall of the query workload.

Supporting top-k queries over distributed collections of schemaless XML data poses two challenges. While XML supports expressive query languages such as XPath and XQuery, these languages require schema knowledge so as to write an appropriate query which may not be available in distributed systems with autonomous and dynamic sources. Thus, there is a need for approximate query processing. Furthermore, retrieving the top-k results incurs large communication and processing cost, since partial result lists from numerous sites need to be combined and ranked to assembly the top-k answers. To address both of these issues, we present an approach for approximate XPath processing over distributed collections of XML data based on a clustered path index, where data is grouped based on structural information. Our method gradually generalizes a query by applying a set of structural transformations to it and the retrieved results are ranked based on the edit distance between two path expressions. A compact indexing data structure is used to reduce the index construction cost. Our experimental results show that our approach significantly reduces the communication cost for retrieving the top-k results, while maintaining a low construction cost for the clustered index.

Peer-to-peer(p2p) systems over an efficient means of data sharing among a dynamically changing set of a large number of a tonomous nodes.Each node in a p2p system is connected with a small number of other nodes thus creating an overlay network of nodes. A query posed at a node is routed through the overlay network towards nodes hosting data items that satisfy it. In this paper, we consider building overlays that exploit the query workload so that nodes are clustered based on their results to a given query workload. The motivation is to create overlays where nodes that match a large number of similar queries are a fewlinks apart. Query frequency is also taken into account so that popular queries have a greater effect on the formation of the overlay than unpopular ones. We focus on range selection queries and se histograms to estimate the query results of each node. Then, nodes are clustered based on the similarity of their histograms. To this end,we introd ce a workload-aware edit distance metric between histograms that takes into account the query workload. Our experimental results show that workload-aware overlays increase the percentage of query results returned for a given number of nodes visited as compared to both random (i.e., unclustered)overlays and non workload-aware clustered overlays (i.e., overlays that cluster nodes based solely on the nodes' content).