Review Comment:
I see many of my comments addressed and think that in particular the example of the washer-dryer helped to answer most of my questions. Nevertheless I still have two main concerns which stop me from accepting this paper or at least let me hesitate:
1. Your ontology is not clearly defined and needs careful review.
2. Even though you explained the ideas of your non-standard reasoning, I am still not sure what is given/detected and what is derived. So what is “true” here and represents “the world”. I only see equivalence classes and I assume that you can create instances by activating services.
The first point is the most crucial for me and I apologize that I did not pay that much attention to that in my previous review. For the second point I would at least like to get some explanation of what is given and what is derived.
Below my more detailed comments:
Ontology:
- According to your ontology, your data properties swst:followerOf and swst:friendOf have xsd:anyURI as range, but in Figure 3 the values they are actually taking are of type xsd:string, these two types are disjoint. In Figure 7, example 1, these data properties become object properties. Please stay consistent and decide for one range. I would suggest to go for object properties since you want to refer your social objects and not the URIs they have.
- If swst:clientEndpoint and swst:serverEndpoint are sub-properties of iot-lite:endpoint then their range is xsd:anyURI, in Figure 3 these are strings, in Figure 7, example 1, these properties become object properties.
- swst:activationValue has range xsd:string. The values in Figure 4 and 7 are integers.
- swst:likeValue: range xsd:decimal, examples (Figure 5 and 7) xsd:double.
- swst:postedBy, similar problem.
OWL-examples:
As far as I understand, you start with a request, which is simply a declaration of an owl equivalence. In you text you state that the request is what the AS sensed (detects). Even though I think I understand your idea behind the process, it confuses me that you use owl here which already has a fixed meaning. So when something is “sensed” I would expect in owl that an instance of that would be created. Instead, your process searches for a subclass of that request which can also be a union of classes. Once that class is found (or partially found) an instance of that class is created. My problem with the creation of an instance, for example by performing a full close, is that, at least as I would understand the logical representation, it should make the subclass be true. So in your example I would expect that the creation of a Full_Close instance would cause rain or at least the detection of rain (because of the equivalence instead of a subclass relation), but rain is not mentioned anywhere in the text. I guess your argumentation here would be that, as long as we do not know otherwise, we can assume rain and therefore create the instance. Nevertheless, I still have the feeling that you change the semantics of an existing Semantic Web Logic - OWL – which I consider harmful since the Semantic Web is based on mutual agreement. If my understanding is right, maybe you can simply replace “detects” by “covers”? I think that would already help. Is that then still correct in your understanding?
Some more questions here:
- you say that the request causes a Concept Covering process, is such a process always started when we have a request or what triggers the process? How does your system know what request is? Everything which is posted on the wall? Everything which belongs to the class “request”?
- You introduce service requests and service descriptions, there seems to be a fundamental difference between those even though the have a similar form: requests need to be “covered”. It could be good to emphasize that.
- Page 14 left: you say that available resources are represented by there functionalities and I think you mean as in Figures 11 and 12, but before you have some “functionalities” described in Figure 4 which is a different kind of descriptions. Since that caused confusion for me, I would suggest to be a little bit more consistent with your terminology. Are service annotations the same as service descriptions?
More (mostly minor) comments:
- Abstract: “the Semantic Web languages are adopted” → all of them?
- General: I often had the feeling that you used OWL and RDF synonymously, please check carefully when you mean what in the paper
- Page 2: According to the SwoT paradigm, standard technologies were adapted… → Standard technologies were adapted according to…
- Page 6: “since the receiver N_i accepts it, they became able to...” → if the receiver N_i accepts it, they become able to...”
- Page 7, no co-ownership. The example is more for friendship not for a follower relation based on no co-ownership.
- Page 8, Definition of like value: can you add a reference where to find the definition of that CNF-induced norm on concept expressions?
- Page 10: LDP specification only supports… → The LDP specification only ….
|
