tfrancart / sparnatural-iswc-2022 Goto Github PK

----------------------- REVIEW 2 ---------------------
SUBMISSION: 346
TITLE: Sparnatural : a visual knowledge graph exploration tool
AUTHORS: Thomas Francart

----------- Originality and Novelty -----------
SCORE: 2 (marginal)
----------- Relevance to the Conference -----------
SCORE: 5 (perfect match)
----------- Impact -----------
SCORE: 4 (high)
----------- Technical soundness -----------
SCORE: 4 (good)
----------- Clarity and quality of presentation -----------
SCORE: 4 (good)
----------- Overall evaluation and Detailed Review -----------
SCORE: -1 (weak reject)
----- TEXT:
This demo paper presents a very interesting exploration tool for knowledge graphs that makes use a user-friendly query building interface for supporting non-expert users in building and running complex queries.
The paper provides details about the functionality and configurability of the tool, and presents two demo scenarios over two RDF datasets which seem to work properly.

However, the paper does not reference and discuss previous works that have proposed exactly the same visual query building interface, in particular ResearchSpace [1] and A-QuB [2]. Thus, the novelty of the described work seems very limited. The author should discuss the differences of the compared work (and its unique characteristics) in relation to these two very relevant works.

Other comments:

• Figure in page 2 is not readable in print.
• Section 2: "as in he following..."
• Section 3: It would be interesting to include more information about the workshops and the gathered feedback.

[1] Oldman, D. and Tanase, D., 2018, October. Reshaping the knowledge graph by connecting researchers, data and practices in ResearchSpace. In International Semantic Web Conference (pp. 325-340). Springer, Cham.
A deployment for the maritime history domain: http://rs.sealitproject.eu/resource/sealit:Search (check the 'semantic search' component)

[2] Kritsotakis, V., Roussakis, Y., Patkos, T. and Theodoridou, M., 2018, September. Assistive Query Building for Semantic Data. In SEMANTICS Posters&Demos.
https://www.ics.forth.gr/isl/assistive-query-building-semantic-data-qub
----------- Summarize the rationale for accepting/rejecting the paper (pros/cons) -----------
Pros:

• Useful (and configurable) tool for exploring knowledge graphs
• Working demos
  Cons:
• The paper does not reference and discuss previous works that have proposed exactly the same visual query building interface, in particular ResearchSpace [1] and A-QuB [2]. Thus, the novelty of the described work seems very limited.

----------------------- REVIEW 3 ---------------------
SUBMISSION: 346
TITLE: Sparnatural : a visual knowledge graph exploration tool
AUTHORS: Thomas Francart

----------- Originality and Novelty -----------
SCORE: 4 (innovative)
----------- Relevance to the Conference -----------
SCORE: 4 (clearly relevant)
----------- Impact -----------
SCORE: 3 (good)
----------- Technical soundness -----------
SCORE: 2 (poor)
----------- Clarity and quality of presentation -----------
SCORE: 2 (poor)
----------- Overall evaluation and Detailed Review -----------
SCORE: 0 (borderline paper)
----- TEXT:
This demo submission describes Sparnatural, which is a knowledge graph exploration tool. The tool allows end-users to "build" SPARQL queries by a beautiful UI with auto complete features and logical connectors. As a fan of efforts that bridges the bag between our technology and end-users, I really like the direction of this work.

Although interesting, there is a lack of, at least, an overview of the inner workings of the tool. I think the paper would benefit a lot from an architecture figure showcasing the involved entities.

The paper editorial work needs to improve. Several paragraphs are not correctly indented. I suggest the author to reformat the paper with the LNCS latex template, in the case of acceptance. In addition, the paper has many typos, including:

• Abstract: developped, cient-side
• Section 3: misplaced colon (:)
• Section 4: misplaced semi-colon (;)
  ----------- Summarize the rationale for accepting/rejecting the paper (pros/cons) -----------
  Pros
• Nice tool for visually building SPARQL queries
  Cons
• The paper needs a serious editorial review
• There is no discussion or mention about similar tools like RDF Explorer

----------------------- REVIEW 1 ---------------------
SUBMISSION: 346
TITLE: Sparnatural : a visual knowledge graph exploration tool
AUTHORS: Thomas Francart

----------- Originality and Novelty -----------
SCORE: 2 (marginal)
----------- Relevance to the Conference -----------
SCORE: 5 (perfect match)
----------- Impact -----------
SCORE: 4 (high)
----------- Technical soundness -----------
SCORE: 4 (good)
----------- Clarity and quality of presentation -----------
SCORE: 2 (poor)
----------- Overall evaluation and Detailed Review -----------
SCORE: 1 (weak accept)
----- TEXT:
The authors presented a tool that guides users in formulating queries for RDF KGs. The tool provides interfaces for triple patterns that users can fill in with widgets (choosing classes, choosing values, etc.). The use of an "intermediate" representation to facilitate query formulation, which can be tailored to different types of end-users, is compelling (even if I question the appropriateness of OWL for this representation).

The source code and demos have been made available, which is appreciated. I suggest publishing the video using FigShare instead of YouTube for this paper.

The tool seems to work smoothly. I was able to explore the KG with the tool quickly, and I was pleasantly surprised by how quickly I figured out how I could add, remove, and rename columns in the tool.

That said, I couldn't manage to create IN / NOT IN filters. Or the use of VALUES in a FILTER, for that matter. And while the authors have mentioned that UNION is not supported, I have the impression that named graphs are not supported either. The paper did not mention query modifiers, and I could not find how to use DISTINCT, LIMIT, etc. The authors could clarify these details in the paper. Creating graphs and switching languages removes the whole graph, which is frustrating.

I do not see the added values of including references [1-4], as they are not used in the text. This would create room to refer to some related work missing from this poster paper. Examples include SPARQL Playground (Bottoni et al., 2015), Visual SPARQL Builder (https://leipert.github.io/vsb/#, and RDF Explorer (Vargas et al., 2019).

The paper contains many typos, most of which a decent spell checker can spot. The typos do not impede the paper's readability but cannot be published as such. I believe proofreading the paper and addressing the typos in a timely manner is feasible.

Overall, I believe the work presented in this paper is worthy of being presented to the community. The main reasons I am apprehensive of vouching for a strong acceptance are 1) the language (see below for a non-exhaustive list of issues) and 2) the lack of (some) related work.

Comments
Throughout the paper, references are clickable but point to the HTML version of the paper published on GitHub and not the references at the bottom of the document.
The authors can omit the technical detail of CORS-enabled endpoints and the reference to the latter, creating space to report on more exciting aspects of this work.

Abstract

• The paragraph is too long and should not contain paragraphs. Please consider rephrasing the paragraph to highlight the main points. Below are some questions w.r.t. the terminology used in the abstract. They either need to be addressed in the abstract, or the abstract will need to be rewritten so that these are described in the text.
• There's no point in introducing the acronyms ANF and BNF in the paragraph as they are not reused. The text should introduce them. Authors can also omit the acronyms from the title of Section 3; i.e., "Demonstrators" does suffice.
• What is meant by "distance" in "shorten the distance between KGs and users"?
• What is "flexibility" in "the flexibility offered by the graph structure"?

Introduction

• What is meant by "programmatic interfaces"?
• The text mentions screenshots, but there is only one. The text should also refer to Fig. 1 explicitly.
• The tool mimics the structure of a "triple pattern", not a "triple".
• To improve readability, avoid using both "predicate" and "link" in the text. The same comment about using "nodes" and "entities" in the paper.

Section 3

• Did you confront the tool with end-users, or did you confront end-users with the tool? I presume the latter.

Section 4

• Why have the authors proposed using OWL ontologies to configure the tool. Does the tool rely on OWL reasoning to achieve certain things? If yes, what? Why did the authors not consider a vocabulary or even (SHACL) shapes?
• "a single predicate in the search interface can correspond to a property path in the underlying graph structure" -> Do the authors retain the predicate's IRI of the KG, or do they use a new IRI? If you reuse the same IRI, would you not end up in a situation where the predicate has different semantics in the tool? Would that not confuse ontology engineers?
• "The configuration ontology does not need to be exactly the same as the ontology of the KG." Does that imply that it needs to be, to some extent, the same? Or, to some extent, follow the same structure? Please elaborate or qualify?

Section 5

• This section presents a summary rather than a conclusion. I would suggest labeling it as such.

Editorial
Title

• There should not be a space before the colon ':' (This problem appears in multiple places in the text).
• Can you indicate the number of participants and provide figures on who could use it (right away) and those who were confused?

Abstract

• "non-expert" and "nonexpert" are both used throughout the paper. Please choose one. The Oxford Dictionary seems to prefer the latter. Same for "end user" and "end-user".
• "loosing" should be "losing"
• "cient-side" -> "client-side"
• "end user" -> "end-users"
• "developped" -> "developed"
• "has make" -> "has made"
• "with a limited effort" -> "with little effort" presuming that the authors want to state that using the tool required little effort and not that users were limited in the amount of effort they were allowed to invest in the tool.
• "enthusiastics" -> "enthusiastic"
• "offered, but also" -> use "and" instead of "but". Authors may want to consider splitting this sentence into two sentences for clarity.
• "an efficient" -> "efficient"
• "How Sparnatural is configured..." -> The sentence starts with "How" and should thus be a question. Consider rephrasing this sentence.

Introduction

• "culturage" -> "cultural"
• "efforts" -> "effort"

Section 2

• "he following" -> "the following"
• "optionnally" -> "optionally"
• "... type and optionally some ..." -> "... type and, optionally, some ..."
• "one criterion", "two criteria", thus "a criteria" -> "a criterion"

Section 3

• This is not important, but I had the impression BNF stood for
• "allows to explore" -> "allows exploring"
• again, a space before a colon.
• "criterias" -> "criteria" ("criteria" is already plural)

Section 4

• "associate with" instead of "associate to"

Section 5

• "... on the configuration side the ability ..." -> "... on the configuration side, the ability ..."
  ----------- Summarize the rationale for accepting/rejecting the paper (pros/cons) -----------
  Pro:
• a new tool with its source available and demos that work
• a library has adopted the tool
• an excellent match with the conference
  Cons:
• language (but can easily be addressed)
• some related work is missing
• some details need to be clarified