Conference Paper

RanKloud: a scalable ranked query processing framework on hadoop.

DOI: 10.1145/1951365.1951444 Conference: EDBT 2011, 14th International Conference on Extending Database Technology, Uppsala, Sweden, March 21-24, 2011, Proceedings
Source: DBLP

ABSTRACT The popularity of batch-oriented cluster architectures like Hadoop is on the rise. These batch-based systems successfully achieve high degrees of scalability by carefully allocating resources and leveraging opportunities to parallelize basic processing tasks. However, they are known to fall short in certain application domains such as large scale media analysis. In these applications, the utility of a given data element plays a vital role in a particular analysis task, and this utility most often depends on the way the data is collected or interpreted. However, existing batch data processing frameworks do not consider data utility in allocating resources, and hence fail to optimize for ranked/top-k query processing in which the user is interested in obtaining a relatively small subset of the best result instances. A naïve implementation of these operations on an existing system would need to enumerate more candidates than needed, before it can filter out the k best results. We note that such waste can be avoided by utilizing utility-aware task partitioning and resource allocation strategies that can prune unpromising objects from consideration. In this demonstration, we introduce RanKloud, an efficient and scalable utility-aware parallel processing system built for the analysis of large media datasets. RanKloud extends Hadoop's MapReduce paradigm to provide support for ranked query operations, such as k-nearest neighbor and k-closest pair search, skylines, skyline-joins, and top-k join processing.

  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: In this paper we examine the problem of ecient feature evaluation for logistic regression on very large data sets. We present a new forward feature selection heuristic that ranks features by their estimated eect on the resulting model's performance. An approximate optimization, based on backtting, provides a fast and accurate estimate of each new feature's coecient in the logistic regression model. Further, the algorithm is highly scalable by parallelizing simultaneously over both features and records, allowing us to quickly evaluate billions of potential features even for very large data sets.
    Proceedings of SIAM International Conference on Data Mining (SDM'09); 01/2009
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Assume that each object in a database has m grades, or scores, one for each of m attributes. For example, an object can have a color grade, that tells how red it is, and a shape grade, that tells how round it is. For each attribute, there is a sorted list, which lists each object and its grade under that attribute, sorted by grade (highest grade first). Each object is assigned an overall grade, that is obtained by combining the attribute grades using a fixed monotone aggregation function, or combining rule, such as min or average. To determine the top k objects, that is, k objects with the highest overall grades, the naive algorithm must access every object in the database, to find its grade under each attribute. Fagin has given an algorithm (“Fagin's Algorithm”, or FA) that is much more efficient. For some monotone aggregation functions, FA is optimal with high probability in the worst case. We analyze an elegant and remarkably simple algorithm (“the threshold algorithm”, or TA) that is optimal in a much stronger sense than FA. We show that TA is essentially optimal, not just for some monotone aggregation functions, but for all of them, and not just in a high-probability worst-case sense, but over every database. Unlike FA, which requires large buffers (whose size may grow unboundedly as the database size grows), TA requires only a small, constant-size buffer. TA allows early stopping, which yields, in a precise sense, an approximate version of the top k answers. We distinguish two types of access: sorted access (where the middleware system obtains the grade of an object in some sorted list by proceeding through the list sequentially from the top), and random access (where the middleware system requests the grade of object in a list, and obtains it in one step). We consider the scenarios where random access is either impossible, or expensive relative to sorted access, and provide algorithms that are essentially optimal for these cases as well.
    Journal of Computer and System Sciences 01/2001; 66(4-66):614-656. · 1.09 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: This paper explores the problem of computing pairwise sim- ilarity on document collections, focusing on the application of \more like this" queries in the life sciences domain. Three MapReduce algorithms are introduced: one based on brute force, a second where the problem is treated as large-scale ad hoc retrieval, and a third based on the Cartesian product of postings lists. Each algorithm supports one or more approx- imations that trade eectiveness for eciency, the charac- teristics of which are studied experimentally. Results show that the brute force algorithm is the most ecient of the three when exact similarity is desired. However, the other two algorithms support approximations that yield large ef- ciency gains without signicant loss of eectiveness.
    Proceedings of the 32nd Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, SIGIR 2009, Boston, MA, USA, July 19-23, 2009; 01/2009

Full-text (2 Sources)

Available from
Aug 25, 2014