Increasingly modern data science platforms today have non-intrusive and extensible provenance ingestion mechanisms to collect rich provenance and context information, handle modifications to the same file using distinguishable versions, and use graph data models (e.g., property graphs) and query languages (e.g., Cypher) to represent and manipulate the stored provenance/context information. Due to the schema-later nature of the metadata, multiple versions of the same files, and unfamiliar artifacts introduced by team members, the resulting "provenance graphs" are quite verbose and evolving; further, it is very difficult for the users to compose queries and utilize this valuable information just using standard graph query model. In this paper, we propose two high-level graph query operators to address the verboseness and evolving nature of such provenance graphs. First, we introduce a graph segmentation operator, which queries the retrospective provenance between a set of source vertices and a set of destination vertices via flexible boundary criteria to help users get insight about the derivation relationships among those vertices. We show the semantics of such a query in terms of a context-free grammar, and develop efficient algorithms that run orders of magnitude faster than state-of-the-art. Second, we propose a graph summarization operator that combines similar segments together to query prospective provenance of the underlying project. The operator allows tuning the summary by ignoring vertex details and characterizing local structures, and ensures the provenance meaning using path constraints. We show the optimal summary problem is PSPACE-complete and develop effective approximation algorithms. We implement the operators on top of Neo4j, evaluate our query techniques extensively, and show the effectiveness and efficiency of the proposed methods.

Deep learning has improved state-of-the-art results in many important fields, and has been the subject of much research in recent years, leading to the development of several systems for facilitating deep learning. Current systems, however, mainly focus on model building and training phases, while the issues of data management, model sharing, and lifecycle management are largely ignored. Deep learning modeling lifecycle generates a rich set of data artifacts, e.g., learned parameters and training logs, and it comprises of several frequently conducted tasks, e.g., to understand the model behaviors and to try out new models. Dealing with such artifacts and tasks is cumbersome and largely left to the users. This paper describes our vision and implementation of a data and lifecycle management system for deep learning. First, we generalize model exploration and model enumeration queries from commonly conducted tasks by deep learning modelers, and propose a high-level domain specific language (DSL), inspired by SQL, to raise the abstraction level and thereby accelerate the modeling process. To manage the variety of data artifacts, especially the large amount of checkpointed float parameters, we design a novel model versioning system (dlv), and a read-optimized parameter archival storage system (PAS) that minimizes storage footprint and accelerates query workloads with minimal loss of accuracy. PAS archives versioned models using deltas in a multi-resolution fashion by separately storing the less significant bits, and features a novel progressive query (inference) evaluation algorithm. Third, we develop e cient algorithms for archiving versioned models using deltas under co-retrieval constraints. We conduct extensive experiments over several real datasets from computDeep learning has improved state-of-the-art results in many important fields, and has been the subject of much research in recent years, leading to the development of several systems for facilitating deep learning. Current systems, however, mainly focus on model building and training phases, while the issues of data management, model sharing, and lifecycle management are largely ignored. Deep learning modeling lifecycle generates a rich set of data artifacts, e.g., learned parameters and training logs, and it comprises of several frequently conducted tasks, e.g., to understand the model behaviors and to try out new models. Dealing with such artifacts and tasks is cumbersome and largely left to the users. This paper describes our vision and implementation of a data and lifecycle management system for deep learning. First, we generalize model exploration and model enumeration queries from commonly conducted tasks by deep learning modelers, and propose a high-level domain specific language (DSL), inspired by SQL, to raise the abstraction level and thereby accelerate the modeling process. To manage the variety of data artifacts, especially the large amount of checkpointed float parameters, we design a novel model versioning system (dlv), and a read-optimized parameter archival storage system (PAS) that minimizes storage footprint and accelerates query workloads with minimal loss of accuracy. PAS archives versioned models using deltas in a multi-resolution fashion by separately storing the less significant bits, and features a novel progressive query (inference) evaluation algorithm. Third, we develop efficient algorithms for archiving versioned models using deltas under co-retrieval constraints. We conduct extensive experiments over several real datasets from computer vision domain to show the efficiency of the proposed techniques.

Deep learning has improved the state-of-the-art results in many domains, leading to the development of several systems for facilitating deep learning. Current systems, however, mainly focus on model building and training phases, while the issues of lifecycle management are largely ignored. Deep learning modeling lifecycle contains a rich set of artifacts and frequently conducted tasks, dealing with them is cumbersome and left to the users. To address these issues in a comprehensive manner, we demonstrate ModelHub, which includes a novel model versioning system (dlv), a domain-specific language for searching through model space (DQL), and a hosted service (ModelHub).

Social recommendation takes advantage of the influence of social relationships in decision making and the ready availability of social data through social networking systems. Trust relationships in particular can be exploited in such systems for rating prediction and recommendation, which has been shown to have the potential for improving the quality of the recommender and alleviating the issue of data sparsity, cold start, and adversarial attacks. An appropriate trust inference mechanism is necessary in extending the knowledge base of trust opinions and tackling the issue of limited trust information due to connection sparsity of social networks. In this work, we offer a new solution to trust inference in social networks to provide a better knowledge base for trust-aware recommender systems. We propose using a semiring framework as a nonlinear way to combine trust evidences for inferring trust, where trust relationship is model as 2-D vector containing both trust and certainty information. The trust propagation and aggregation rules, as the building blocks of our trust inference scheme, are based upon the properties of trust relationships. In our approach, both trust and distrust (i.e., positive and negative trust) are considered, and opinion conflict resolution is supported. We evaluate the proposed approach on real-world datasets, and show that our trust inference framework has high accuracy, and is capable of handling trust relationship in large networks. The inferred trust relationships can enlarge the knowledge base for trust information and improve the quality of trust-aware recommendation.

Advertising activity on SNS has grown rapidly and is now a billion dollar business. In the SNS advertising model, the SNS serves as the advertising agent, and takes the advantage of network diffusion to attract advertisers and charges for the cascading impressions. The optimal ad allocation task is to choose the ad allocation plan that maximizes the revenue. Each user has different diffusion ability, limited daily impressions and the advertisers have various bidding prices and budget concerns. A feasible plan that obeys the constraints is difficult to find. The solution of this problem lies in the space of N |Ads|× |user|, which makes direct optimization unattractive. In this paper, we study SNS advertising business models, formulate the SNS ad allocation problem and show their connections with hyperbolic embedding. We develop a new embedding algorithm Hyper Cube Map that allows for dimension reduction. Our proposed method reduces the dimensionality of the original problem significantly, runs two to four orders of magnitude faster, and reaches 95% of the optimum.

In modern data science problems, techniques for extracting value from big data require performing large-scale optimization over heterogenous, irregularly structured data. Much of this data is best represented as multi-relational graphs, making vertex-programming abstractions such as those of Pregel and GraphLab ideal fits for modern large-scale data analysis. In this paper, we describe a vertex-programming implementation of a popular consensus optimization technique known as the alternating direction method of multipliers (ADMM) [1]. ADMM consensus optimization allows the elegant solution of complex objectives such as inference in rich probabilistic models. We also introduce a novel hypergraph partitioning technique that improves over the state-of-the-art vertex programming framework and significantly reduces the communication cost by reducing the number of replicated nodes by an order of magnitude. We implement our algorithm in GraphLab and measure scaling performance on a variety of realistic bipartite graphs and a large synthetic voter-opinion analysis application. We show a 50% improvement in running time over the current GraphLab partitioning scheme.

There is a growing interest in methods for analyzing data describing networks of all types, including biological, physical, social, and scientific collaboration networks. Typically the data describing these networks is observational, and thus noisy and incomplete; it is often at the wrong level of fidelity and abstraction for meaningful data analysis. This demonstration presents GrDB, a system that enables data analysts to write declarative programs to specify and combine different network data cleaning tasks, visualize the output, and engage in the process of decision review and correction if necessary. The declarative interface of GrDB makes it very easy to quickly write analysis tasks and execute them over data, while the visual component facilitates debugging the program and performing fine grained corrections.