Periodic-frequent pattern mining is a vital knowledge discovery technique that aims to find all regularly occurring patterns in a temporal database. Previous studies focused on developing CPU-centric algorithms by disregarding the speedups offered by the GPUs. Furthermore, existing GPU-based frequent pattern mining algorithms cannot be employed to find periodic-frequent patterns because they ignore the items' temporal occurrence information in the database, and the multi-threaded sum-reduction technique cannot be employed to determine the periodicity of a pattern in a database. With this motivation, this paper proposes an efficient GPU-accelerated depth-first search algorithm, GPU Periodic Frequent-Miner (gPF-Miner), to find the desired patterns. Our algorithm employs a novel flattened array structure to effectively record the temporal occurrence information of every item in a database. Our algorithm also introduces a new multi-threaded parallelization technique to calculate the support and periodicity of a pattern in a GPU. This technique’s best and worst-case time complexities are O(1) and O(n), where n represents the data size. Experimental results demonstrate that gPF-Miner outperforms the existing CPU-based and naive GPU-based algorithms by a vast margin.

A geo-referenced time series database represents the data generated by a set of fixed locations (or spatial items) observing a particular phenomenon over time. This data hides valuable information that can help users progress in their social and economic lives. This paper presents a new model of Geo-referenced Periodic-Frequent Patterns (GPFPs) that might be in these databases. A GPFP is a set of frequently occurring items close to each other and seen in a database at regular intervals. Three constraints have been used to figure out how interesting a pattern is in a geo-referenced time series database: maximum distance (maxDist), minimum support (minSup), and maximum periodicity (maxPer). The maxDist controls how far apart the items in a pattern can be. The minimum number of times a pattern must appear in the data is controlled by the minSup. Lastly, the maxPer variable specifies how many times a pattern must repeat before it is considered periodic in the data. Each pattern that satisfies these three requirements will be returned. An effective method known as the Geo-referenced Periodic-Frequent Pattern-Miner (GPFP-Miner) has been proposed to discover all GPFPs included inside a geo-referenced time series database. GPFP-Miner uses an innovative, smart depth-first search approach to uncover required patterns efficiently. The findings of the experiments support the contention that the proposed algorithm is effective. In addition, we present two case studies in which we utilise our methodology to extract meaningful information from databases pertaining to air pollution and traffic congestion.

Partial periodic pattern mining is an important model in data mining with many real-world applications. However, this model’s successful industrial application was hindered by the problem of combinatorial explosion of patterns, which involves generating too many patterns, most of which might be redundant or uninteresting to the user. Furthermore, the problem of combinatorial explosion increases the memory, runtime, and the energy requirements of a mining algorithm. This paper aims to tackle this challenging problem by proposing a novel model of maximal partial periodic pattern that may exist in a database. We also present a new tree structure and a pattern-growth algorithm, called Maximal Partial Periodic Pattern-growth (max3P-growth), to find all desired patterns effectively. Experimental results demonstrate that the proposed model prunes many redundant patterns, and the max3P-growth is efficient and scalable. Finally, we show the usefulness of our model with a case study on traffic congestion analytics.

Finding Spatial High Utility Itemsets (SHUIs) in a spatiotemporal database is a challenging problem of great importance in many real-world applications. Most previous works focused on the sequential discovery of SHUIs in a database running on a single machine. Consequently, these works are not suitable for big data (or cloud-based) applications as they suffer from the scalability and fault tolerant problems. This paper proposes several novel pruning techniques to reduce the search space and present a more flexible distributed algorithm to find all desired itemsets from the database using Spark in-memory computing architecture. Our algorithm inherits several advantages of Spark, including low communication cost, fault tolerance, and high scalability. Experimental results demonstrate that the proposed algorithm has good scalability and performance on very large databases. Finally, we present a real-world navigation application in which SHUIs generated from the traffic congestion data have been employed to recommend alternative routes to the users.

Periodic-frequent pattern mining (PFPM) is an important data mining model having many real-world applications. However, the successful industrial application of this model has been hindered by the problem of combinatorial explosion of patterns, that is the generation of too many redundant patterns, most of which may be useless to the user. To address this problem, this paper proposes a novel model of maximal periodic- frequent pattern that may exist in a temporal database. A new pattern-growth algorithm, called Maximum Periodic-Frequent Pattern-growth (maxPFP-growth), has also been introduced to efficiently find all desired patterns in the data. Experimental results demonstrate that maxPFP-growth is not only memory and runtime efficient, but also highly scalable as well. The usefulness of our model has also been demonstrated with a case study on traffic congestion analytics.

Most keyword searches target precision for finding the most relevant document. However some target recall, finding all relevant documents. Our system supports high recall searches that return hundreds or thousands of relevant results. In particular, it provides a visualization that shows the distribution of search results relative to the distribution of items for the entire corpus. Such relative distributional features include over and under representation, clusters and outliers. The contribution of this paper is efficient visualisation, that is, how to provide the best relative distribution view for a given data cube size. This requirement is translated to: for which limited size meta-data summary cube are search results disambiguated the most in our relative distribution view. We identify metrics and several algorithms for such a summary cube selection.

We demonstrate a system that integrates a novel OLAP component with a keyword search engine, to support querying over sparse and ragged corpus data. The key contribution of our system is the integration of dynamically selected point sets such as search results with OLAP querying over aggregated data. During the demonstration, participants will be able to enter a keyword search; observe the returned list of result files; observe distributional features such as outliers and clusters of results in the corpus in multiple dimension views; and select and partition corpus slices in the OLAP component to narrow search results. Participants will be able to experience not just the individual querying features of our system, but the way that they work together to facilitate smooth interaction sequences that combine OLAP and keyword search querying.