Social Internet of Things for Domotics: a Knowledge-based Approach over LDP-CoAP

Tracking #: 1636-2848

Authors: 
Michele Ruta
Floriano Scioscia
Giuseppe Loseto
Filippo Gramegna
Saverio Ieva
Agnese Pinto
Eugenio Di Sciascio

Responsible editor: 
Guest Editors ST Built Environment 2017

Submission type: 
Full Paper
Abstract: 
Ambient Intelligence aims at simplifying the interaction of a user with her surrounding context minimizing the effort needed to increase comfort and assistance. Nevertheless, especially in built and structured environments, current technologies and market solutions are often far to provide the required levels of automation, coordination and adaptivity of the ambient. This paper proposes a novel semantic-based framework complying the emerging Social Internet of Things paradigm. Infrastructured spaces can be intended as populated by device agents organized in social networks, interacting autonomously and sharing information, cooperating and orchestrating resources. A service-oriented architecture allows collaborative dissemination, discovery and composition of service/resource descriptions. The Semantic Web languages are adopted as linguistic layer and mobile-oriented implementations of non-monotonic inferences for semantic matchmaking are used to give decision capabilities to software agents. Finally, the Linked Data Platform (LDP) over the Constrained Application Protocol (CoAP) provides the knowledge organization and sharing infrastructure underpinning social object interactions. The framework has been implemented and tested in a home automation prototype integrating several communication protocols and off-the-shelf devices. Experiments advocate the effectiveness of the approach.
Full PDF Version: 
Tags: 
Reviewed

Decision/Status: 
Major Revision

Solicited Reviews:
Click to Expand/Collapse
Review #1
By Charbel El Kaed submitted on 14/Jun/2017
Suggestion:
Major Revision
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.

The paper presents a Social Internet Of Things in the context of the Home and Building Automation. The approach proposes an ambient intelligence scenario where IoT sensors/actuators/devices cooperate with each others to re-adapt to an internal reconfiguration or an outside event such as an intrusion detection. The authors propose a vision where the IoT "things" interact in a human-like social network. They have the possibility to send a "friend request" or just be a "follower". In addition, the things have a "wall" where they can publish information or updates. Such information is either triggered by a change in their context, after an internal change or from their other IoT devices "friends" and/or "followers".
In this Service Oriented Architecture social IoT network, the "things" can publish their services and modifications of the supported services and environment, while other devices can listen and subscribe.
The approach relies on the semantic web standards and proposes a heterogeneous architecture where the things can have different capability and only a set of nodes can embed reasoning to match services definitions while basic things supports only limited discovery capabilities.
The approach was validated over LDP-CoAP. The evaluation was performed with a set of 8 KNX devices and the smart node was developed on three platforms, a RPI, Intel Edison, and UDoo.

There are some outstanding concerns regarding the review criteria:

1- The paper presents the idea of social internet of things which is original in the sense of augmenting the things with a social network aspect. However, the authors cite that their approach is similar to Voutyras et al but does not present the differentiating elements. Voutyras et al also proposes a social network of objects. Therefore, it is hard to assess the originality of the approach.

2- The paper presents a social internet of things, however, the aspects covered in the paper are obvious and the interesting aspects are still missing, which at least should be pointed out if it is not yet covered:
-The concepts Abduction and Contraction, are not clear, an example can help the reader better understand the approach.
-It is not clear how the friendship or follow decision will be made, on which criteria the nodes will base their decision. Can a device refuses a friend request or a follow request
-How can a device prove its identity to other potential friends? Cybersecurity risks are not covered in this work but at least the risk should be pointed out in the paper
-In the collaborative adaptivity concept, an example will help the reader understand the concept.
- An explanation about the semantic service description and the matchmaking procedure will help the reader understands better the approach.
- How to avoid flooding of other nodes with information. For example Intrusion is a high risk event and should be propagated. It is not clear how the propagation will scale?
- The evaluation was performed with a set of 8 KNX devices and the smart node was developed on three platforms, a RPI, Intel Edison, and UDoo.
I believe 8 devices is a pretty small number.
Based on these aspects, it is an interesting approach however, still in its early days.

3- As for the quality of writing, as suggested in the comments below and before, the paper is confusing. It fails to present a clear understanding of the approach where the case study and the technical details are mixed.

For these reason, I believe that the paper is not ready for publication and that authors should be allowed to improve, answer, and clarify the remaining points.

In the following, notes and remarks:
[Section 2. state of the art]
a. The authors mention the paper of Voutyras et al which proposes a social network of objects, it is not clear how their work differentiate.
b. The state of the art is missing previous work related to reasoning and semantic alignment in the SOA domain.

[Section 3. A social network for smart linked objects]
c. The concepts Abduction and Contraction, are not clear, an example can help the reader better understand the approach.
d. It is not clear how the friendship or follow decision will be made, on which criteria the nodes will base their decision. Can a device refuses a friend request or a follow request
e. How can a device prove its identity to other potential friends? Cybersecurity risks are not covered in this work but at least the risk should be pointed out in the paper
f. In the collaborative adaptivity concept, an example will help the reader understand the concept.
g. sub-section 3.2 LDP-CoAP is too technical, better to be moved to the implementation section.
h. Moving the case study or at least giving examples in the beginning will help the reader to understand how the approach is applicable in the Home automation domain instead of waiting until section 4.

[Section 4]
i. Figure 8 is 2 pages away from the reference.
j. There are formalization of ontology definition without any caption. They should be references as Listings.
k. Figures 9 and 10 are too far as well and better declare them as listings.
l. An explanation about the semantic service description and the matchmaking procedure will help the reader understands better the approach.
m. Figure 7 too technical, maybe moved to the implementation section.
n. It is not clear how the "like" feature is used and how it is updated. Seems to be used to stop information propagation.
o. Is every update on the wall shared among 'friends' and 'followers', it is not clear how the sharing of information is handled, by priority or other
p. How to avoid flooding of other nodes with information. For example Intrusion is a high risk event and should be propagated. It is not clear how the propagation will scale?

[5. Evaluation] page 14 is full of charts and a table without proper commit or description.

Review #2
Anonymous submitted on 20/Jun/2017
Suggestion:
Major Revision
Review Comment:

The present paper discusses a possible implementation of a Semantic Social Internet of Things and provides a case study for that implementation set up in a smart home environment.

The paper has a very interesting topic and I like the idea of modelling different kinds of relationships between devices which can support each other in different tasks. However, the description of the details of the implementation and its underlying logical concepts is rather poor. That makes it difficult to form an opinion about the approach itself. I therefore recommend to clarify the explanations, in particular the logic. More detailed:

Logic: I would like to see the logic treated with more care. Some examples:
- In figure 3, I see that the device is subject to sor:friendOf and soic:follows. From this use of predicates and their ontologies follows that the device is an foaf:Person and an foaf:OnlineAccount at the same time. Maybe that peculiarity is intended, but in such case I would like to see an explanation for that in the text.
- Figure 3 and text: you indicate in your text the property iot-lite:endpoint is used for the server and swst:clientEndpoint for the client. But you also state that the last predicate is a subproperty of the first. You say that you involve reasoning. How does your engine know which of the two iot-lite:endpoint links to the actual server?
- How are figures 3-6 logically connected? Where are the links?

Semantic matchmaking (Chapter 3): This part seems to be the most interesting to me and it is really unfortunate that I do not fully grasp what you are doing here. This section needs clarification. I furthermore recommend to add concrete examples (triples). In particular:
- Do you first produce the deductive closure of your ontology and data and then do some kind of query (request)?
- How does a request look like? Can it contain variables?
- Concept contraction: How do conflicting elements look like? Is it always that easy to find and resolve the conflict (in other contexts that can be a challenge)? Here I would like to see an example.
- Concept Abduction: When do you do concept abduction instead of concept covering?

Collaborative adaptivity:
- Is “the configuration requested by the node” the same as a request?
- Like value: I still don’t know what this “like value” is. Can you either give a proper definition or show a better example?

Semantic Web of Social Things Ontology: I think readers will also be interested in your ontology and, if they like it, would even consider to reuse it. Can you make the ontology publicly available?

Case study: My questions here are closely related to the previous ones: I don’t see how the request works and would like you to clarify that. You write:
“The AS verifies if the services of the connected objects , SC_1 and L_1 in our example were modified and ….”
How exactly is this check caused? I do not see the connection between the request and the information in figure 9 and 10.

Evaluation: You explain what you did in your evaluation, but it is still not clear for me, as a potential user of your implementation, what I gain from your evaluation? Why did you chose these different settings? What do you want to show?

Minor remarks and typos:
- page 1: surroundings objects → surrounding objects
- page 3: An in-depth analysis […] is in [41], … → can be found in, is presented in
- page 3: … swrl rules were used to implement temporal and extra-logical constraints. → I think it is just the wording, but I doubt that you can do extra-logical things with a rather weak rule language like swrl (but I am curious to learn).
- page 4: From a linguistic point of view this paper refers to […] ALN Desscription logics → I would not call the choice of the logic a linguistic point of view
- page 4: f_c → f_i I think it is rather unusual to refer to the index of f_1, f_2, … as f_c since c normally stands for a constant, I would rather use I, j, or k
- page 5: don’t → do not
- page 5: on the its wall → its wall
- page 11: it should notice a significant variation → a significant variation can be noticed
- page 13: Table 4 highlights as only → that only

Review #3
By Fulvio Corno submitted on 23/Jun/2017
Suggestion:
Accept
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.

The paper presents a complete and comprehensive framework for social collaboration among smart object using ontologies and linked-data representations.
The proposed framework builds upon existing standards and ontologies, and extends previous work from the same authors.
The paper re-interpreters the social concepts of friend, follower, post, annotation, and gives them new meaning in the context of the collaboration of smart objects (that behave like a multi agent system).
While the solution architecture draws from the MAS domain, the authors adapted it to the IoT world by adopting suitable communication protocols, by managing the limited computational resources, and by adopting taxonomies of device and service types.

The paper may be improved in the following aspects:

1. how does the framework address the specific needs of IoT systems? you take into account communication protocols and computing power (both important issues, but low-level), but are there any specific functional or architectural aspects, specific to IoT, that influenced your proposed framework? In other words, why is this a framework for IoT, and not simply for MAS?

2. the mapping of smart objects' features and behaviors to semantic operation is not always clear from the text. In describing the semantic operations (e.g. in section 3), you should refer to their usage (ore necessity) in the domain of IoT. For example, "a node [...] requiring adaptation" refers to a specific semantic condition; what does it corresponds to, from the smart object point of view? what are the IoT-level functionality that is enabled/supported by this adaptation procedure? This happens often in the paper, where you are taking for granted a mapping between "IoT features" and "semantic computations" that is not explicit in the paper.
The same goes for the relationship of "posts" with "services".
One other concept that is very hardly understood is the "likeness".

3. Concerning the class modeling (e.g. Fig 11), you should state what models are applicable in general (e.g., modeling the security aspects in such a way that it applies to any smart objects that covers that functionality), and what models are specific for the use case (i.e., they should be modeled individually for each smart home). Are the presented rules generic or are they fitted to H1 and H2?

The paper is well organized and clearly written, and it refers to the relevant literature.


Comments