Enterprise customers increasingly require greater flexibility in the way they access and process their Big Data while at the same time they continue to request advanced analytics and access to diverse data sources. Yet customers also still require the robustness of enterprise class analytics for their mission-critical data. In this paper, we present our initial efforts toward a solution that satisfies the above requirements by integrating the HPE Vertica enterprise database with Apache Spark's open source big data computation engine. In particular, it enables fast, reliable transferring of data between Vertica and Spark; and deploying Machine Learning models created by Spark into Vertica for predictive analytics on Vertica data. This integration provides a fabric on which our customers get the best of both worlds: it extends Vertica's extensive SQL analytics capabilities with Spark's machine learning library (MLlib), giving Vertica users access to a wide range of ML functions; it also enables customers to leverage Spark as an advanced ETL engine for all data that require the guarantees offered by Vertica.

Due to R's popularity as a data-mining tool, many distributed systems expose an R-based API to users who need to build a distributed application in R. As a result, data scientists have to learn to use different interfaces such as RHadoop, SparkR, Revolution R's ScaleR, and HPE's Distributed R. Unfortunately, these interfaces are custom, non-standard, and difficult to learn. Not surprisingly, R applications written in one framework do not work in another, and each backend infrastructure has spent redundant effort in implementing distributed machine learning algorithms. Working with the members of R-core, we have created ddR (Distributed Data structures in R), a unified system that works across different distributed frameworks. In ddR, we introduce a novel programming primitive called dmapply that executes functions on distributed data structures. The dmapply primitive encapsulates different computation patterns: from function and data broadcast to pair-wise communication. We show that dmapply is powerful enough to express algorithms that fit the statistical query model, which includes many popular machine learning algorithms, as well as applications written in MapReduce. We have integrated ddR with many backends, such as R's single-node parallel framework, multi-node SNOW framework, Spark, and HPE Distributed R, with few or no modifications to any of these systems. We have also implemented multiple machine learning algorithms which are not only portable across different distributed systems, but also have performance comparable to the "native" implementations on the backends. We believe that ddR will standardize distributed computing in R, just like the SQL interface has standardized how relational data is manipulated.

Due to R's popularity as a data-mining tool, many distributed systems expose an R-based API to users who need to build a distributed application in R. As a result, data scientists have to learn to use different interfaces such as RHadoop, SparkR, Revolution R's ScaleR, and HPE's Distributed R. Unfortunately, these interfaces are custom, non-standard, and difficult to learn. Not surprisingly, R applications written in one framework do not work in another, and each backend infrastructure has spent redundant effort in implementing distributed machine learning algorithms. Working with the members of R-core, we have created ddR (Distributed Data structures in R), a unified system that works across different distributed frameworks. In ddR, we introduce a novel programming primitive called dmapply that executes functions on distributed data structures. The dmapply primitive encapsulates different computation patterns: from function and data broadcast to pair-wise communication. We show that dmapply is powerful enough to express algorithms that fit the statistical query model, which includes many popular machine learning algorithms, as well as applications written in MapReduce. We have integrated ddR with many backends, such as R's single-node parallel framework, multi-node SNOW framework, Spark, and HPE Distributed R, with few or no modifications to any of these systems. We have also implemented multiple machine learning algorithms which are not only portable across different distributed systems, but also have performance comparable to the "native" implementations on the backends. We believe that ddR will standardize distributed computing in R, just like the SQL interface has standardized how relational data is manipulated.

Users often write reviews on different themes involving linguistic structures with complex sentiments. The sentiment polarity of a word can be different across themes. Moreover, contextual valence shifters may change sentiment polarity depending on the contexts that they appear in. Both challenges cannot be modeled effectively and explicitly in traditional sentiment analysis. Studying both phenomena requires multi-theme sentiment analysis at the word level, which is very interesting but significantly more challenging than overall polarity classification. To simultaneously resolve the multi-theme and sentiment shifting problems, we propose a data-driven framework to enable both capabilities: (1) polarity predictions of the same word in reviews of different themes, and (2) discovery and quantification of contextual valence shifters. The framework formulates multi-theme sentiment by factorizing the review sentiments with theme/word embeddings and then derives the shifter effect learning problem as a logistic regression. The improvement of sentiment polarity classification accuracy demonstrates not only the importance of multi-theme and sentiment shifting, but also effectiveness of our framework. Human evaluations and case studies further show the success of multi-theme word sentiment predictions and automatic effect quantification of contextual valence shifters.

As the website is a primary customer touch-point, millions are spent to gather web data about customer visits. Sadly, the trove of data and corresponding analytics have not lived up to the promise. Current marketing practice relies on ambiguous summary statistics or small-sample usability studies. Idiosyncratic browsing and low conversion (browser-to-buyer) make modeling hard. In this paper, we model browsing patterns (sequence of clicks) via Markov chain theory to predict users' propensity to buy within a session. We focus on model complexity, imputing missing values, data augmentation, and other attendant issues that impact performance. The paper addresses the following aspects; (1) Determine appropriate order of the Markov chain (assess the influence of prior history in prediction), (2) Impute missing transitions by exploiting the inherent link structure in the page sequences, (3) predict the likelihood of a purchase based on variable-length page sequences, and (4) Augment the training set of buyers (which is typically very small: 2% by viewing the page transitions as a graph and exploiting its link structure to improve performance. The cocktail of solutions address important issues in practical digital marketing. Extensive analysis of data applied to a large commercial web-site shows that Markov chain based classifiers are useful predictors of user intent.

Big data applications are growing rapidly as more sensors are connected to the internet and gathering business critical information for processing. Imaging sensors are an important type of sensors for collecting images and video data, and are widely deployed on smartphones, video surveillance networks, and robots. Traditional databases are designed to ingest and search structured information. The analysis of unstructured information such as images and videos is often done separately. In this paper, we describe a big data system with deep integration of face analysis and recognition in images and videos. We show how we can utilize the built-in parallelization in the Vertica database system to accelerate feature computation and search. We also show an example application of the system for face re-identification and face search in images and videos.

Graph analytics is becoming increasingly popular, with a number of new applications and systems developed in the past few years. In this paper, we study Vertica relational database as a platform for graph analytics. We show that vertex-centric graph analysis can be translated to SQL queries, typically involving table scans and joins, and that modern column-oriented databases are very well suited to running such queries. Furthermore, we show how developers can trade memory footprint for significantly reduced I/O costs in Vertica. We present an experimental evaluation of the Vertica relational database system on a variety of graph analytics, including iterative analysis, a combination of graph and relational analyses, and more complex 1-hop neighborhood graph analytics, showing that it is competitive to two popular vertex-centric graph analytics systems, namely Giraph and GraphLab.

Combining data warehousing and stream processing technologies has great potential in offering low-latency data-intensive analytics. Unfortunately, such convergence has not been properly addressed so far. The current generation of stream processing systems is in general built separately from the data warehouse and query engine, which can cause significant overhead in data access and data movement, and is unable to take advantage of the functionalities already offered by the existing data warehouse systems.In this work we tackle some hard problems in integrating stream analytics capability into the existing query engine. We define an extended SQL query model that unifies queries over both static relations and dynamic streaming data, and develop techniques to extend query engines to support the unified model. We propose the cut-and-rewind query execution model to allow a query with full SQL expressive power to be applied to stream data by converting the latter into a sequence of "chunks", and executing the query over each chunk sequentially, but without shutting the query instance down between chunks for continuously maintaining the application context across the execution cycles as required by sliding-window operators. We also propose the cycle-based transaction model to support Continuous Querying with Continuous Persisting (CQCP) with cycle-based isolation and visibility.We have prototyped our approach by extending the PostgreSQL. This work has resulted in a new kind of tightly integrated, highly efficient system with the advanced stream processing capability as well as the full DBMS functionality. We demonstrate the system with the popular Linear Road benchmark, and report the performance. By leveraging the matured code base of a query engine to the maximal extent, we can significantly reduce the engineering investment needed for developing the streaming technology. Providing this capability on proprietary parallel analytics engine is work in progress.

Support transaction property and fault-tolerance is the key to applying stream processing to industry-scale applications; however the corresponding latency overhead must be minimized for accommodating real-time analytics. This issue has been studied in various contexts. In this work we develop the backtrack failure recovery mechanism to allow a task to roll forward without waiting for acknowledgement from its downstream target tasks in the failure-free case, but to request its upstream source tasks to resend the missing tuples only during failure recovery which is the rare case thus has limited impact on the overall performance. For further reduced latency we extend our solution in another dimension by applying the notion of optimistic checkpointing to stream processing, and propose the Continued stream processing with Window - based Checkpoint and Recovery (CWCR) approach, allowing a task to emit results tuple by tuple continuously but checkpoint in batch, and acknowledge, only once per window (e.g. time window). We also tackle the hard problems found in implementing a transactional layer on-top of an existing stream processing platform. We have implemented the proposed mechanisms on Fontainebleau, the distributed stream analytics infrastructure we built on top of the open-sourced Storm platform. Our experiment results reveal the novelty of the proposed technologies and the feasibility to support fault-tolerance with minimal latency overhead for real-time stream processing.

A typical predictive analytics workflow will pre-process data in a database, transfer the resulting data to an external statistical tool such as R, create machine learning models in R, and then apply the model on newly arriving data. Today, this workflow is slow and cumbersome. Extracting data from databases, using ODBC connectors, can take hours on multi-gigabyte datasets. Building models on single-threaded R does not scale. Finally, it is nearly impossible to use R or other common tools, to apply models on terabytes of newly arriving data. We solve all the above challenges by integrating HP Vertica with Distributed R, a distributed framework for R. This paper presents the design of a high performance data transfer mechanism, new data-structures in Distributed R to maintain data locality with database table segments, and extensions to Vertica for saving and deploying R models. Our experiments show that data transfers from Vertica are 6x faster than using ODBC connections. Even complex predictive analysis on 100s of gigabytes of database tables can complete in minutes, and is as fast as in-memory systems like Spark running directly on a distributed file system.

Adverse reactions to drugs are a major public health care issue. Currently, the Food and Drug Administration (FDA) publishes quarterly reports that typically contain on the order of 200,000 adverse incidents. In such numerous incidents, low frequency events that are clinically highly significant often remain undetected. In this paper, we introduce a visual analytics system to solve this problem using (1) high scalable interfaces for analyzing correlations between a number of complex variables (e.g., drug and reaction); (2) enhanced statistical computations and interactive relevance filters to quickly identify significant events including those with a low frequency; and (3) a tight integration of expert knowledge for detecting and validating adverse drug events. We applied these techniques to the FDA Adverse Event Reporting System and were able to identify important adverse drug events, such as the known association of the drug Avandia with myocardial infarction and Seroquel with diabetes mellitus, as well as low frequency events such as the association of Boniva with femur fracture. In our evaluation, we found over 90% of the adverse drug events that were published in the Institute for Safe Medication Practices (ISMP) reports from 2009 to 2012. In addition, our domain expert was able to identify some previously unknown adverse drug events.