Survey of Models and Architectures to Ensure Linked Data Access

Tracking #: 3076-4290

Mahamadou Toure
Kaladzavi Guidedi
Fabien Gandon
Moussa Lo
Christophe Guéret

Responsible editor: 
Ruben Verborgh

Submission type: 
Survey Article
Mobile Access to the Web of Data is currently a real challenge in developing countries, mainly characterized by limited Internet connectivity and high penetration of mobile devices with the limited resources (such as cache and memory). In this paper, we survey and compare proposed solutions (such as models and architectures) that could contribute to solving this problem of mobile access to the Web of Data with intermittent Internet access. These solutions are discussed in relation to the underlying network architectures and data models considered. We present a conceptual study of peer-to-peer solutions based on gossip protocols dedicated to design the connected overlay networks. In addition, we provide a detailed analysis of data replication systems generally designed to ensure the local availability of data on the system. We conclude with some recommendations to achieve a connected architecture that provides mobile contributors with local access to the Web of data.
Full PDF Version: 


Solicited Reviews:
Click to Expand/Collapse
Review #1
By Edgard Marx submitted on 13/Jul/2022
Minor Revision
Review Comment:

In general, the paper addressed most of the issues raised in the first review.
However, I think the organization should be revised.
Here I give some suggestions.


You can mention that you provide a “detailed” analysis of client-server and “data replication …”


One of the contributions is a comprehensive overview of the different methods to consume RDF data which is not mentioned.

#Section "4.3. Other relevant work on peer-to-peer architecture"

I suggest renaming this section to: RDF-based peer-to-peer architectures.

Please remove KBox from there as well as Linked Data fragments.
Linked Data Fragments is a client/server architecture that transfers part of the processing to the client, it is not peer-to-peer.
KBox is a distributed and decentralized RDF data dependency managed system that pushes the query processing to the client using replicas of the graph.
KBox should be moved to section 5.2. Graph replication while LDF, in my opinion, deserves its section, maybe Hybrid Architectures.
Also please make sure that you emphasize that KBox storage, CPU, and runtime among others efficiency is based on the fact that a single client is executing the query while client-server architectures centralized data access on the server.
Further, KBox efficiency came with the cost of transferring the graph to the client which may not be desirable in many scenarios i.e. mobile platforms.

#Section "5. RDF Graph Sharing and collaborative modification"

I suggest moving section 5.1.1 to the beginning of the Survey, i.e Section 3. Preliminaries.

Section 5.1.2 and 5.1.3 should be moved to a new section i.e 6. RDF Client-Server architectures and systems.


Please use capital letters when referring to Semantic Web.

Review #2
By Carsten Keßler submitted on 14/Jul/2022
Review Comment:

In my original review, I was a bit torn between "reject" and "major revision" and went with the latter to give the authors a chance to explain why this topic warrants a survey article. Unfortunately, the revision does not achieve this goal. I am still not convinced that this is useful as introductory text to the topic, simply because the topic itself is very fuzzy and the article covers a hodgepodge of all kinds of different aspects of data access (any kind of data, not just linked data!). The conclusion reads a bit like the introduction to a PhD thesis, and if this is indeed the case, much of the work that has gone into this manuscript can hopefully still be used for the related work chapter of the thesis.

Review #3
Anonymous submitted on 21/Sep/2022
Review Comment:

This manuscript aims to present solutions for mobile access to the Web of Data in developing countries, where connectivity is limited. Unfortunately, it does not succeed at this goal: the paper meanders from a far-fetched motivating case, to a description of P2P (not relevant to the SWJ audience), to a short overview of distributed RDF approaches, but fails to bring the point home on how this concretely helps in low-connectivity environments. I am missing new insights, and am worried about the correctness of some of the descriptions and discussions in there (as discussed below).

As such, I do not think this paper is a good fit for the journal.

- The introduction is quite chaotic. It jumps from very basic Linked Data to connectivity problems on the African content to quite specific details of peer-to-peer networks, then back to high-level solutions and back to peer-to-peer. The claimed contributions are then around gossip protocols on a high level, and then specifically RDF sharing systems.

- The long introduction about linked data is not appropriate for SWJ; our readers know these concepts already, and at a deeper level than is in the introduction.

- Reference 1: who is J. Timothy?

- Reference 7 seems to be a very obscure work to back up a broad statement on P2P; I don't think it is appropriate here.

- The motivation case is quite contrived; such activities are not performed even when there is connectivity. There is no indication whatsoever that anyone would want to engage in such activities.

- Additionally, it seems like a very small amount of data, that would be a couple of kilobytes at most.

- And there is no indication further in the manuscript that the motivational case is actually addressed appropriately.

- The authors write "Research on RDF data exchange on peer-to-peer networks has made great progress.", but this statement is not backed up; the closest reference [14] is not related to RDF.

- This section is devoted to an explanation of gossip protocols, which already exists elsewhere (notably [14]).

- Pages 4–9 not related to RDF at all. While announced in the intro, they are not relevant to SWJ readers.

- The discussion on RDF-aware systems is quite haphazard and unstructured; it it not helpful to our audience.

- KBox is a technique for sending the entire knowledge graph to the client; its relevance here is unclear.


- The RDF explanation is too basic for the Semantic Web Journal, and not all information is correct or appropriately phrased.
– "RDF Schema is a [special] vocabulary"
- N3 is not a representation of RDF
– Jena, Sesame, and Virtuoso are entirely irrelevant here
- HDT is not a solution to limitations of RDF
- …

- There is some value in 5.2.2 where different replication approaches are considered; similar for 5.3.3. Those are the most valuable parts of the paper. However, they are only a limited part of the whole.