Search Results for "data collection algorithm"

The DISTOD data profiling algorithm is a distributed algorithm to discover bidirectional order dependencies (in set-based form) from relational data. DISTOD is based on the single-threaded FASTOD-BID algorithm [1], but DISTOD scales elastically to many machines outperforming FASTOD-BID by up to orders of magnitude. Bidirectional order dependencies (bODs) capture order relationships between lists of attributes in a relational table. They can express that, for example, sorting books...

Open data often comes with little or no metadata. You will profile a large collection of open data sets and derive metadata that can be used for data discovery, querying, and identification of data quality problems. For each column, identify and summarize the semantic types present in the column. These can be generic types (e.g., city, state) or collection-specific types (NYU school names, NYC agency). For each semantic type T identified, enumerate all the values encountered for T in all...

Linked Open Data (LOD) has emerged as one of the largest collection of interlinked datasets on the web. Benefiting from this mine of data requires the existence of descriptive information about each dataset in the accompanying metadata. Such meta information is currently very limited to few data portals where they are usually provided manually thus giving little or bad quality insights. To address this issue, we propose a scalable automatic approach for extracting, validating and generating...