Understanding Data Science Lifecycle Provenance via Graph Segmentation and Summarization

Abstract

Along with the prosperous data science activities, the importance of provenance during data science project lifecycle is recognized and discussed in recent data science systems research. Increasingly modern data science platforms today have nonintrusive 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 "provenance graph" is verbose and evolving, and hard to understand; using standard graph query model, it is difficult to compose queries and utilize this valuable information.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. The operators are implemented on top of a property graph backend. We evaluate our query methods extensively and show the effectiveness and efficiency of the proposed methods.

A Step toward Next-Generation Advancements in the Internet of Things Technologies

“…Data science helps to process and supports the derivation of valuable data [ 59 ]. Figure 8 illustrates the data science lifecycle [ 12 , 60 ]. The lifecycle is based on the seven iterative steps such as data cleaning, data exploration, and data mining, etc.…”

Production Machine Learning Pipelines: Empirical Analysis and Optimization Opportunities

“…The issues of inter-connectedness and size of provenance graphs have similarly emerged in different domains, wherein techniques such as user views, segmentation, and aggregation have been explored to transform the graphs to usable or interpretable ones [9,17,37,38]. We adopt a similar approach but we leverage the semantics of production-ML operators and connections between them.…”

“…Previous work has even led to the standardization of provenance representations for workflows in the form of graphs [26,39,40]. Other research has proposed various ways to explore and analyze such provenance graphs, e.g., visualization [14], reachability query support [13], support for user-defined views [17], segmentation and summarization [9,10,37]. Our work introduces a framework to segment ML provenance graphs and demonstrates how this segmentation leads to further analysis and optimizations for ML pipelines.…”

An Interactive and Predictive Pre-diagnostic Model for Healthcare based on Data Provenance

“…Furthermore, given information about the place data originated from, how they come in their present states, and who or what acted on them helps users to establish trust in the data. Provenance can show resources and relations that have affected the construction of the output data and are commonly expressed as directed graphs (digraphs) [17]. The primary aim of the W3C standardized provenance is to enable the extensive publication and exchange of provenance over the web [18].…”