No abstract

We study the problem of identifying discriminative features in Big Data arising from heterogeneous sensors. We highlight the heterogeneity in sensor data from engineering applications and the challenges involved in automatically extracting only the most interesting features from large datasets. We formulate this problem as that of classification of multivariate time series and design shapelet-based algorithms for this task. We design a novel approach, called Shapelet Forests (SF), which combines shapelet extraction with feature selection. We evaluate our proposed method with other approaches for mining shapelets from multivariate time series using data from real-world engineering applications. Quantitative analysis of the experiments shows that SF performs better than the baseline approaches and achieves high classification accuracy. In addition, the method enables identification of noisy sensors from multivariate data and discounts their use for classification.

Malicious software ('malware')

Several operations in the Exploration and Production (E&P) sector are event-driven in nature and are supported by specialized systems and applications. Narrow focus of applications results in application silos that restrict the information sharing across verticals, which is a critical requirement for coordinated cross-functional efforts. Effective response to events warrants due emphasis on an integration strategy that facilitates desired information flow across verticals. Event-driven methods can be used to make strategic asset management decisions across silos in real-time, thus reducing response time and costs while improving asset performance.Complex event processing is an emerging research area that involves detecting complex events, processing the events, deciding actions for each event and notifying the relevant personnel about the event. In the E&P sector, the adoption of CEP and messaging-based systems in conjunction with semantic methods can facilitate components of the oilfield to communicate in real-time across different software platforms. Such an approach helps not only in detecting complex events across various sources, but also in processing them and deciding the actions to be taken, with the help of a knowledge base -thereby reducing information overload.Consider a typical application scenario -a pump failure event in an oilfield, which should elicit response not only by the pump operator but also by the maintenance engineers, production managers, reservoir engineers and other involved personnel. A proactive event-driven system enables quick detection of the failure across heterogeneous data sources and takes corrective actions while notifying the appropriate personnel. This facilitates effective communication across the teams and software systems involved.We propose a semantic complex event processing architecture for the digital oilfield that facilitates enterprise information integration. We delineate an illustrative use case of such integration for production optimization. Value propositions of the proposed framework include efficient interaction patterns, reduction in data seeking efforts, faster response times, building of consistent best practices and management by exception.