Towards a Semantic Modelling of Profiling Data in Industry 4.0

Tracking #: 2223-3436

Authors: 
Imran Jami
Shaukat Wasi
Siraj Munir

Responsible editor: 
Guest Editors SemWeb of Things for Industry 4.0 - 2019

Submission type: 
Full Paper
Abstract: 
In this paper we present a framework for semantic representation of profiling information of workers and employees in Industry 4.0 setting using the IoT devices. With the integration of Internet of Things with Big Data and Semantic Web, we provide effective retrieval of profiling information of actors in real-time within the industry. It aims at supporting intelligent queries about the workers working in the Industry by using temporal knowledge graph generated from the multi attributed logs identified from the surveillance video. The paper facilitates integration of the profiling knowledge graph with existing industry knowledge graph for unified view of profiling log with workers information. Decentralized decision making in real time can be made with this system which is one of the core areas of Industry 4.0. Knowledge Graph is proposed as generic for reusability in any industries and enterprise for profiling purpose. The results show the highlights of the queries about working in the industry without the involvement of any resource. The paper is concluded with shortcomings in this work along with the solution.
Full PDF Version: 
Tags: 
Reviewed

Decision/Status: 
Reject

Solicited Reviews:
Click to Expand/Collapse
Review #1
Anonymous submitted on 07/Oct/2019
Suggestion:
Major Revision
Review Comment:

The paper presented a framework for semantic representation of profiling information of workers and employees in Industry4.0. The paper is presentation is good but the novelty of this paper is not clearly presented along with other implementation details.

Following points need to be addressed:
1. Introduction section information flow needs to get revised to clearly state the gap or why this work is unique?. References are required at places such as "some researcher (line 30)".
ii. Ambiguous use of the word the Author in section 2.
iii. The author talk on Image segmentation in the literature review. It is not clear how this is related to this work. If so, need to be clarified.
iv. Auther used CNN for facial recognition, however, there is not any description on CNN model training, test, accuracy. As the accuracy of facial recognition is directly related to the performance of this model So, an analysis will be better to analyze the performance of this model based on the accuracy of facial recognition. accuracy.
v. A simple graph model is presented with location and employee. A detailed semantic representation will be better to include.
vi Auther talk about large scale data sets (what scale? description will be better). How information is generated is described in the Result and discussion section.

Review #2
Anonymous submitted on 28/Oct/2019
Suggestion:
Reject
Review Comment:

This manuscript was submitted as 'full paper' and should be reviewed along the usual dimensions for research contributions which include (1) originality, (2) significance of the results, and (3) quality of writing.

This paper proposes a semantic model to store profile data of workers in a factory. The proposed approach uses different sensing and face recognition technologies to create user movement log and based on user profiles and their activities the proposed system is capable of answering a few user-related queries. Authors claim that their approach will lead towards a knowledge graph for workers and their movement which can be integrated into the overall knowledge base of the factory.

I have the following serious concerns regarding the work proposed in this paper;

* The proposed solution of having a semantic model for users does not appear to be a valid reason for using semantics. In the discussed example, the model is able to store only a few basic information regarding user/worker profiles and their movement logs. It is absolutely not clear why semantics are needed here and where they are used in their model?
* Figure 1 presented as a data model for employee profiling, gives a stepwise information flow. A variety of components in this model are presented, with absolute no information on why and where these components are used by the proposed solution. For example, it is not clear what is meant by Semantic Engine and Rule-based? Are these components performing some tasks? If so, what is their role?
* A set of 5 queries given in the introduction consists of very basic queries, which could be answered by any system which has stored users profiles and their movement log. It is not clear, what authors meant by semantic queries here? Why we need a graph database and what was not possible with a simple join between user-profiles and their activity logs?
* Role of IoT devices and camera is also dubious, authors mentioned a few technologies like sensors and camera feed based face recognition, but it is not clear if they were used or authors are assuming that this is done by someone else using state of the art solutions?
* Semantic Engine is showed in Figure 7, I have serious concerns on calling it a Semantic Query Engine as it appears it is just a list of pre-populated questions, which can be translated into queries.
* A few example questions and their equivalent query translation is given in examples, but not clear if this translation is manual or it can be done for any generic queries.
* Serious concerns around privacy and security will arise with any solution which allows tracking users' location during the work hours within a factory.
* Overall, I believe this paper is simply providing a very high-level idea of building users' profiles and keeping their movement logs. Using this information, authors are trying to show that a set of questions can be posed. I am not sure, if there is any novelty in this work, as storing employees' profiles and then location logs can be as simple a storing a few tables in the database and putting primary foreign key relations. This paper leaves a lot of basic questions, such as
Why semantics is needed here? Why using a Graph database was a necessity? What authors meant by the knowledge graph? Is this a working system? What size data was used for testing and validation e.g. how many workers profile was created and how long their movement logs were stored?

I have strong reservation on the novelty of the proposed solution, and authors need to show that the semantics-based solution is bringing what kind of novel aspects which are beyond state of the art.

In the current state of the paper, which requires a lot of writing quality improvement and very little (if any) contribution beyond state of the art solutions. Hence, I recommend that this paper is not in a state that it could be accepted and even any revisions require substantial changes.