Downloads: ViStream

ViStream 1.0 Version [Released Aug 20, 2013]

Features:

Our system allows the user to perform visual analytic tasks on streaming data. The system

  • identifies and extracts patterns of interesting (e.g. clusters) [1,2,3],
  • summarizes discovered patterns at multiple resolutions and saves them in a historical archive [4],
  • matches the live patterns against the historical archive [4],
  • monitors evolution of patterns (e.g., continuity, dissolve, split, and merge of clusters) over time [5],
  • visualizes the patterns of interest and supports interactions in an interactive user interface [6].

Binary Releases:

  • Download the zip file and follow the instructions to install ViStream into your system.

Source Releases:

Tutorials:

Videos:

  • Please enable caption for more information.

    • Screenshots:

      Query Management Interface Query Management
    • 1. Select Data Source, currently we integrated three types of data into our system.
    • 2. With or without tracking evolution pattern [5].
    • 3. Config the parameters for the sliding window and the density based clustering algorithm.
    • 4. Enable or disable the summerization and matching feature [4]
    • 5. Save the query plan, that will be run on the query engine. The query plans are placed in a temporary location by windows if you run as normal user.
    • 6. Manage the existing query plans.
      Query Engine Interface Query Engine
    • Select a query plan created before and add it to the query engine. The detail of the query plan is displayed when the "show query plan" button is pressed.
      Visul Engine Interface Visual Engine
    • The change of patterns are displayed on the screen, and the user can watch the evolving patterns in real time. The display includes cluster structure evloving view on the top of the screen, and the visual representation of the data points of each cluster in a Parallel Coordinates view (what is this view?) on the bottom of the screen.
      Pattern Matching Request Pattern Matching Request
    • 1. Streaming window visualizes the extracted patterns in real time;
    • 2. The user can interacts with a snap shot of the streaming window. The user initiates a matching request by clicking on one colored bubble (indicates a cluster) in the current snap shot. Our system finds the matched clusters in the history archive and visualize them in a Cluster Match Display;
    • 3. The cluster Match Display shows the found matched pattern. The time the pattern occurs are represented by a window ID. The detail of the matched clusters are visualized by a Parrallel Coordinates view;

    Trouble Shooting:

    • If the system crashes, please go to "start" and type in "eventvwr" in the "search programs and files" box and hit "enter".
    • On the left tree view panel of the "Event Viewer" window unfold "Windows Logs" and click on "Application" tab, look up the application error in the top right panel and target the errors associated with either "XmdvTool.exe", "power_stream_query_engine.exe", or "power_stream_source_engine.exe"
    • Double click on related error entries, copy the detailed error message and send the error message to us: xmdv@cs.wpi.edu.

    Data Sources:

    Parse Streaming Data (Assuming in XML format):

    Since different data streams have different data formats, our system supports parsing data streams in the XML format into the format our engine requires. You only need to know XQuery to specify this conversion for your own data set.

    References:

    • [1]. Di Yang, Elke A. Rundensteiner, Matthew O. Ward: Mining neighbor-based patterns in data streams. Inf. Syst. 38(3): 331-350 (2013) (Paper link)
    • [2]. Di Yang, Elke A. Rundensteiner, Matthew O. Ward: Shared execution strategy for neighbor-based pattern mining requests over streaming windows. ACM Trans. Database Syst. 37(1): 5 (2012) (Download paper)
    • [3]. Di Yang, Elke A. Rundensteiner, Matthew O. Ward: A Shared Execution Strategy for Multiple Pattern Mining Requests over Streaming Data. PVLDB 2(1): 874-885 (2009) (Download paper)
    • [4]. Di Yang, Elke A. Rundensteiner, Matthew O. Ward: Summarization and Matching of Density-Based Clusters in Streaming Environments. PVLDB 5(2): 121-132 (2011) (Download paper)
    • [5]. Yang, Di, Zhenyu Guo, Elke A. Rundensteiner, and Matthew O. Ward. "CLUES: a unified framework supporting interactive exploration of density-based clusters in streams." In Proceedings of the 20th ACM CIKM, 815-824. (2011) (Download paper)
    • [6] Di Yang and Kaiyu Zhao and Hanyuan Lu and Maryam Hasan and Elke A. Rundensteiner and Matthew O. Ward, "Mining and Linking Complex Patterns across Live Data Streams and Stream Archives", VLDB (2013), to appear (Download paper Download poster)