Essential Issues to Consider for a Manufacturing Data Query System Based on Graph

Abstract

Manufacturing industry data are distributed, heterogeneous and numerous, resulting in different challenges including the fast, exhaustive and relevant querying of data. In order to provide an innovative answer to this challenge, the authors consider an information retrieval system based on a graph database. In this paper, the authors focus on determining the essential functions to consider in this context. The authors define a three-step methodology using root causes analysis and resolution. This methodology is then applied to a data set and queries representative of an industrial use case. As a result, the authors list four major issues to consider and discuss their potential resolutions.

i-Dataquest: A heterogeneous information retrieval tool using data graph for the manufacturing industry

“…Because we aim to meet the challenges listed by (Kim, et al, 2020), we are looking for an information retrieval system solution based on a graph constructed from the enterprise's heterogeneous data exploiting both the textual and tabular content of text and image documents as well as database records while semantically extending the terms of the query. While commercial solutions for information retrieval in companies are numerous but they do not offer a solution that meets all the criteria sought for the manufacturing context.…”

“…Related work in section 2 shows the relevance of using graph-oriented databases to represent and search for relational information in different fields. In particular, in the context of information retrieval systems using a graph of data, the literature lacks a solution considering both structured and unstructured data while providing an answer to the issues listed by (Kim, et al, 2020) including the treatment of syntactic specificities, the semantic research and finally the display of particularly relevant results. Concerning the specific issue of semantic extension, the state of the art has highlighted the possibilities offered by lexical networks, which can take the form of a multilingual knowledge graph such as ConceptNet.…”

“…In our case, we choose β= 1 judging that it is just as important to obtain all relevant results as to display only relevant results. Finally, each missing or excessive result is analysed according to a root cause analysis and then classified using an Ishikawa 13 (Barsalou, 2014) method that allowed the identification of large families of anomalies (Kim, et al, 2020).…”

“…The work of (Kim, et al, 2020) depicted the main milestones in order to define an information retrieval system for manufacturing industry based on pre-processed data under a graph-oriented data model. Indeed, this system must consider several main challenges including the expansion of keywords to semantically close terms, the treatment of syntactic heterogeneity of source data and the display of particularly relevant results.…”

“…To improve the capacities of the manufacturing industry to exploit its data while taking into account the previous main challenges, a framework for a graph-based information retrieval system has been proposed (Kim, et al, 2020). The aim of this paper is to present the i-dataquest proposal that fits the framework that has been enriched, implemented and evaluated.…”