A Semantic Approach to Model Multimedia Information and Social Networks for Cultural Heritage

Tracking #: 2347-3560

Authors: 
Antonio Rinaldi
Cristiano Russo

Responsible editor: 
Special Issue Cultural Heritage 2019

Submission type: 
Full Paper
Abstract: 
The social aspect of information has a crucial role in our everyday life. In this context, the representation and management of Online Social Networks (OSNs) represent a new challenge in the research community. In particular, the use of heterogeneous data needs an extension of OSNs to Multimedia Social Networks (MSNs). In this paper we propose a general high-level model to represent and manage MSNs. Our approach is based on a property graph represented by a hypergraph structure due to the intrinsic multidimensional nature of social networks and semantic relations to better represent the networks contents using semantic web vision. In addition, the use of the proposed graph structure allows to discriminate different levels of knowledge analyzing the relationships defined between nodes of the same or different type. Moreover, the introduction of low-level multimedia features and a formalization of their semantic meanings give a more comprehensive view of the social network structure and content. Using this approach we call the represented network Multimedia Semantic Social Networks (MS2N). The proposed data model, based on a top ontological model for knowledge representation, could be useful for several applications. We also propose a case study on cultural heritage domain.
Full PDF Version: 
Tags: 
Reviewed

Decision/Status: 
Reject

Solicited Reviews:
Click to Expand/Collapse
Review #1
Anonymous submitted on 17/Mar/2020
Suggestion:
Major Revision
Review Comment:

This paper presents a model to handle information of multimedia based social media that relies on graphs (equivalent to ontologies). The authors use this model in a case study where the user is recommended to multimedia based on past preferences.

However, the model seems to me simple and not quite original. Moreover, what really troubles me is the use of low level features/descriptors in the experiment, which has not be included in the model. This means that it is not clear what it actually helps.

The experiment also are not so clear. Initially, the authors present examples of usage of the proposed model which are very different from the case study in the experimental section. Also, the location information used in the experiments to augment to accuracy of the recommendation is not clearly connected to the proposed model.

Overall, I think that the paper needs a major revision, where in the revised manuscript the originality of the model, the problem that the paper attacks and the experiments demonstrating its merits must be presented much more clear. Experiments will be also needed but I cannot propose what experiments before I see clearly which problem the authors aim to solve.

Review #2
Anonymous submitted on 18/Mar/2020
Suggestion:
Reject
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 an ontology model for Multimedia Semantic Social Networks. Such networks have become quite popular on the Web; the topic is important.

In section 1, an introduction to the area is presented. However, I could not find out precisely what is the actual research problem addressed by the paper, but substantial improvements in the paper are still needed before it would ready for SWJ.

Next, related work is discussed. Quite a few papers are referred to, but it is not clear how they relate to the paper at hand. Some specific features of the proposes model are listed in the end of the section, but what are their contributions? This remains unclear.

The model is presented in section 3. More explanations about why these modelling choices were made would be helpful. Also, the explanation is essentially a set of definitions and examples, but the actual model remains vague. How does the model contribute to related graph models is not clear.

A case study and experimental implementation is presented in next (section 4). Again, I found the explanations vague and could not really figure out what kind system actually and been created, for what purpose, and what conclusions can be drawn from the experiments. Were the test users happy with the system? What are its contributions w.r.t. related systems? Handling heterogeneous data is nice but not a novelty. The system has not been evaluated extensively but this remain future work as the authors also say in conclusions.

In short, the originality of the paper is not clear at the current state (1), significance of the results has not really been shown. The quality of the writing is good from a linguistic point of view, but the explanations and the model remain in many cases vague. Lots of work seems to be behind the system and the implementation, but substantial changes are in my mind still needed in the manuscript before it is suitable for the SWJ.

Review #3
Anonymous submitted on 11/Apr/2020
Suggestion:
Reject
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 proposes a graph-based data model for Multimedia Social Networks, namely social networks that enable users to share multimedia content. A case study from the cultural heritage domain is used to demonstrate and evaluate the proposed model.

Although it presents some nice ideas, the paper has, in my opinion, several flaws, the most important of which are that the proposed model is not properly motivated, is not clearly presented and is not sufficiently evaluated.

Regarding its motivation, the paper does not justify the definition of a new graph-based model instead of an ontology, which is the most common data modelling approach in this field. Ontology languages provide features such as unique URIs and inference support, which the proposed model does not seem to support. I would expect the authors to explain why these features may not be important, or how they can be supported by the proposed model, or what other features, which are not supported by ontologies but are supported by the proposed model, may be more important in this domain.

Regarding the presentation of the model, many of the design choices are either not clearly justified or not well explained. For example, it is not clear at all how weights are useful and how they are actually used. It is also unclear how the metrics in Section 3.2 are relevant, and how they are computed. The case study does not make things clearer as it does not use any weights or any of these metrics. On the other hand, it adds more confusion, as it discusses some low-level multimedia features, which is not clear at all if and how they can be modelled by the proposed model.

The evaluation is also very unclear. It is not clear at all what is evaluated and how. The figures do not add any clarity as their labels are very generic and the Y-axes are not labelled. A standard evaluation approach for data modelling would include aspects as accuracy, completeness, clarity, etc., which are not used at all in this paper.

For all these reasons, but also because cultural heritage is only used as a case study and is not the main focus of this research, I recommend the rejection of the paper.