Data Series Management: The Road to Big Sequence Analytics

Massive data series collections are becoming a reality for virtually every scientific and social domain, and have to be processed and analyzed, in order to extract useful knowledge. Current data series management solutions are ad hoc, requiring huge investments in time and effort, and duplication of effort across different teams. Systems like relational databases, Column Stores, and Array Databases are not a suitable solution either, since none of these systems offers native support for data series. Our vision is to design and develop a general-purpose Data Series Management System, able to cope with big data series, that is, very large and fast-changing collections of data series, which can be heterogeneous (i.e., originate from disparate domains and thus exhibit very different characteristics), and which can have uncertainty in their values (e.g., due to inherent errors in the measurements). Just like databases abstracted the relational data management problem and offered a black box solution that is now omnipresent, the proposed system will allow analysts that are not experts in data series management, as well as common users, to tap in the goldmine of the massive and ever-growing data series collections they (already) have.