An SQWRL-Based Method for Assessing Regulatory Compliance in the Pharmaceutical Industry

Round 1

Reviewer 1 Report

I have read with a lot of interest the work entitled "An SQWRL based method for assessing Regulatory Compliance in the Pharmaceutical Industry". From the Semantic Web perspective, this work proposes to use description logics implemented in SWRL and included in a given ontology to validate a knowledge graph reusing that ontology. Parsed using SQWRL queries, SWRL rules allow to answer various competency questions by contrast to shape-based validation methods such as ShEx. From the pharmaceutical industry perspective, it is interesting to implement several constraints for the regulation of medicine production. Although this work has multiple important results that can advance knowledge-based industry management in the medical context, it has several limitations that need to be solved prior to publication:

1. The title needs to be changed to "An SQWRL-based method for assessing regulatory compliance in the pharmaceutical industry".

2. The pharmaceutical side of the topic has only been emphasized in the Introduction and Literature Review. The use and development of Data Integrity Ontology should be motivated from the side of medicines industry. The discussion should also be expanded.

3. Figure 3 only featured relational properties. However, there are also non-relational properties that can be relevant and that can assigned SWRL rules. One of this properties is "hasName" identified in Rule R1. Probably, Figure 3 and SWRL rule description should be expanded and even explained from the side of pharmaceutical industry.

4. The Conclusion should be expanded to include future directions for this research work for the side of semantic web technology and pharmaceutical industry.

As a result, I propose that this work will be accepted for publication after these minor revisions are applied.

Author Response

1. The title needs to be changed to "An SQWRL-based method for assessing regulatory compliance in the pharmaceutical industry".

 

The title of the manuscript has been changed as suggested.

 

2. The pharmaceutical side of the topic has only been emphasized in the Introduction and Literature Review. The use and development of Data Integrity Ontology should be motivated from the side of medicines industry. The discussion should also be expanded.

Regarding  Data Integrity Ontology from the side of pharmaceuticals industry, related content has been added on both, subsection 3.3 and section 5 (Discussion section), as suggested.

 

3. Figure 3 only featured relational properties. However, there are also non-relational properties that can be relevant and that can assigned SWRL rules. One of this properties is "hasName" identified in Rule R1. Probably, Figure 3 and SWRL rule description should be expanded and even explained from the side of pharmaceutical industry.

 

Regarding Figure 3, related explanation content has been added in subsection 3.3. Regarding SWRL rule description related content per each rule has been added in subsection 4.1.

 

4. The Conclusion should be expanded to include future directions for this research work for the side of semantic web technology and pharmaceutical industry.

 

The Conclusion section has been enriched with future directions for this current research work as suggested.

Reviewer 2 Report

- Most references are relatively old, It can be understandable in the introduction section, but the related work part needs more up-to-date references

- Paper is very well written, clear description for all concepts and methodologies

 

 

Author Response

Thank you for your comment. As has mentioned in the manuscript research attempts regarding regulatory compliance in pharmaceutical domain is quite scarce. Indeed there should be much more related work in literature. I have reasons to believe that this particular area has many opportunities to offer for research novelties. Moreover, we have included our most updated research work in reference 34 where the current work has been based on. 

Reviewer 3 Report

The paper targets a vital problem for assessing Regulatory Compliance in the Pharmaceutical Industry. The manuscript uses the SQWRL method for assessing Regulatory Compliance in the Pharmaceutical Industry.

In my judgment, the proposed strategy verifies the structural modality with the framed architecture.

However, the mentioned concerns must be addressed for further clarification in terms of revision.

 

1.      The abstract part was not framed in a convincible manner. Better collate the exact framings related to the problem in a suitable manner.

2.      When introducing any new term, it must be expanded. The terms mentioned in the abstract such as ALCOA, SWRL, and SQWRL, must be with a full abbreviation. This makes the reader more comfortable understanding the terminology.

3.      The methodology construction provided in this manuscript is validated as per the architecture. However, I could not see any formations related to Pharmaceutical Industry which are validated for the same. This is a serious drawback of this paper.

4.      The software architecture provided in section 3.4 is not validated w.r.t the application of the Pharmaceutical Industry. It’s better to provide a suitable description validated for the Pharma industry in a separate section.

5.      On the other hand, the authors have simply presented the query description in section 4.2. How are the queries validated w.r.t industry application? This part is missing. Better clarify on the same.

6.      The result section presented in this paper is related to the data assessment part. How are these applicable and resolve the issues of the Pharma industry? Moreover, no comparison has been made by showing the efficiency of the same. In my opinion, the result section is not showing a promising outcome. Better restructure the result section w.r.t the application of the pharma industry and with required results.

Author Response

1. The abstract part was not framed in a convincible manner. Better collate the exact framings related to the problem in a suitable manner.

 

The abstract part has been restructured so as to be  related to the problem in a more suitable manner, as suggested.

 

2. When introducing any new term, it must be expanded. The terms mentioned in the abstract such as ALCOA, SWRL, and SQWRL, must be with a full abbreviation. This makes the reader more comfortable understanding the terminology.

 

Full description has been added for the abbreviations SWRL, and SQWRL in Introduction section. However, regarding ALCOA term, full abbreviation of all of its 9 principles (including the 4 principles of ALCOA+ as well), had been already placed in the original manuscript version.

 

3. The methodology construction provided in this manuscript is validated as per the architecture. However, I could not see any formations related to Pharmaceutical Industry which are validated for the same. This is a serious drawback of this paper.

 

Methodology construction in relation to Pharmaceutical Industry has been enriched with related content in subsection 3.4.

 

4. The software architecture provided in section 3.4 is not validated w.r.t the application of the Pharmaceutical Industry. It’s better to provide a suitable description validated for the Pharma industry in a separate section.

 

Subsection 3.4 has been enriched with related content and an example has been added as well, but not in a separate section, providing a more suitable description validated for the Pharma industry.

 

5. On the other hand, the authors have simply presented the query description in section 4.2. How are the queries validated w.r.t industry application? This part is missing. Better clarify on the same.

 

With respect to each SQWRL query description additional content w.r.t industry application has been added in subsection 4.2.

 

6. The result section presented in this paper is related to the data assessment part. How are these applicable and resolve the issues of the Pharma industry? Moreover, no comparison has been made by showing the efficiency of the same. In my opinion, the result section is not showing a promising outcome. Better restructure the result section w.r.t the application of the pharma industry and with required results.

 

Result section (section 5), has been restructured with additional explanatory content and Figure 6, focused on the efficiency of the assessment prototype, on the grounds of Pharma industry issues, and its forthcoming anticipations.

Round 2

Reviewer 3 Report

The revised version is not formulated as per the comments raised.

The comments  (3,4,5, and 6) mentioned in the previous draft have not been addressed.

It is suggested to address the same.

Author Response

Comment 3. The methodology construction provided in this manuscript is validated as per the architecture. However, I could not see any formations related to Pharmaceutical Industry which are validated for the same. This is a serious drawback of this paper.

Comment 4. The software architecture provided in section 3.4 is not validated w.r.t the application of the Pharmaceutical Industry. It’s better to provide a suitable description validated for the Pharma industry in a separate section.

Response:

To address these comments, a new subsection (i.e., the Subsection 3.3) has been added in which a detailed description of the SPuMoNI system is provided and validated for the pharmaceutical industry. This subsection (Subsection 3.3) presents various pharmaceutical data formations and corresponding dataflows, originated from various heterogeneous data sources, such as sensors, PLCs, production devices, operator staff, equipment alarms, as these are shown in Figure 1. The new subsection provides a description of the SPuMoNI pharmaceutical industry system which retrieves all these raw data from various heterogeneous data sources. Raw data are preprocessed by SWRL rules and they are further refined into a more qualitative information by SQWRL queries, before entering into the ALCOA assessment module (Figure 1) of the SPuMoNI system, for yielding the final decision of whether a specific batch record of the pharma plant production is ALCOA compliant or non-compliant.

Comment 5. On the other hand, the authors have simply presented the query description in section 4.2. How are the queries validated w.r.t industry application? This part is missing. Better clarify on the same.

Response:

Subsection 4.2 has been enriched with relevant content aiming to provide a detailed justification of how the SQWRL queries are validated w.r.t their industrial application. For each particular SQWRL query, a relevant description has been added which presents how the query detects the violation of a corresponding ALCOA principle in the context of the industrial application of the SPuMoNI system.

Comment 6. The result section presented in this paper is related to the data assessment part. How are these applicable and resolve the issues of the Pharma industry? Moreover, no comparison has been made by showing the efficiency of the same. In my opinion, the result section is not showing a promising outcome. Better restructure the result section w.r.t the application of the pharma industry and with required results.

Response:

Section 5 has been enriched with new relevant content that demonstrates that the data assessment prototype of the SPuMoNI system has been applied and validated on some real world pharma manufacturing data, generated in the context of the actual pharmaceutical processes which take place in the pharmaceutical process plant of the pharmaceutical industrial partner involved in the SPuMoNI project, with the aim to resolve actual ALCOA assessment issues raising in this pharma industry. Compared to the previous paper version, the new version of the manuscript has been enhanced with additional content and figures in order to provide a more comprehensive presentation regarding the applied data assessment process. For example, in Section 5 it is shown that the data assessment prototype of the SPuMoNI system measures the non-violation success percentage per each ALCOA principle, as seen analytically in the JSON printout of Figure 6. The system detects non-compliant production batches and writes down the exact type of failure per each ALCOA principle violation. Then, the system calculates the percentage of succeeded, non-violated batches, given the total number of batches.

Please also consider that currently, the DIOnt ontology has been tested on a small amount of batch data with the aim to analyze its computation performance. Due to the fact that the ontology consists of a great number of classes and properties, which are required for fully describing the pharma industry domain (Figures 2a-d), the run time of the required computations to perform ALCOA compliance analysis would not always be polynomial in the size of the data input. Our future intension is to improve DIOnt computation performance by considering an enhanced version of the ontology, which will be applicable to larger data inputs.