A Dialogue with Linked Data: Voice-based Access to Market Data in the Sahel

Tracking #: 459-1636

Authors: 
Victor de Boer
Nana Baah Gyan
Anna Bon
Wendelien Tuyp
Chris van Aart
Hans Akkermans

Responsible editor: 
Guest Editors Semantic Web For All

Submission type: 
Tool/System Report
Abstract: 
In this paper, we present our ongoing efforts to bring the benefits of the Web of Data to rural communities in the Sahel region. These efforts center around RadioMarch\'e, a market information system (MIS) which can be accessed using first-generation mobile phones. We argue that linking the locally produced and consumed data to (external) Linked Data sources will increase its value. We describe how we have made RadioMarch\'e data accessible as Linked Open Data and present a prototype demonstrator that provides voice-based access to this linked market data. Through this interface, the Linked Data can be accessed using first generation mobile phones. As such, these are first steps towards opening the Web of Data to local users that do not have appropriate hardware to produce and consume Linked Data.
Full PDF Version: 
Tags: 
Reviewed

Decision/Status: 
Minor Revision

Solicited Reviews:
Click to Expand/Collapse
Review #1
By Ivana Marenzi submitted on 28/Apr/2013
Suggestion:
Minor Revision
Review Comment:

The paper describes how the market information system RadioMarche´ data have been made accessible as Linked Open Data, and presents a prototype demonstrator that provides voice-based access to this linked market data. The paper is short (8 pages) and well written. The introduction clearly describes the motivation as well as the contributions of the work. Examples of related works on voice technologies, including technical details, are given as preliminary background.
A clear description of the reference tool RadioMarche´ is provided and both capabilities and limitations are mentioned. The relevance of the system is well described, as well as the impact of exposing locally produced data in developing regions as linked data. The effectiveness in distributing market information was confirmed by the results of the pilot phase, on the other hand, as a drawback, the limitation of the supply chain could not cope with the resulting large demand of products. For this reason the RadioMarche´ system is currently not operational.
In section 4 the Linked Data approach is well introduced and clearly motivated. The process of collecting and exposing data in different languages is detailed and well supported by the examples in the pictures. Figures are clear and well described. The sequence and the numbering need to be fixed though (Figure 3 comes before Figure 2)

The Linked Market Data Demostrator though, seems to be still in a preliminary version. While RadioMarche´ can be used in several languages, the Linked Data demonstrator is available only in English and Dutch. The authors write that one advantage of Linked Data is that “it is well suited to deal with multiple languages”, and that “for the RadioMarche´ system a great number of audio recordings are available”; then the authors should explain why they limited the prototype to English and Dutch, and they did not include the local languages Bambara and Bomu already, since the pilot was run in the specific region. They should also describe how difficult would be to actually include more languages, and give more examples to describe in more depth how additional Linked Data could be integrated and used.

A further proof reading is suggested to fix a few typos (e.g. at page 4 to to GeoNames, and llanguages, in Figure 4.3…at page 5 a in the PHP document…)
References – At 6, 12 and 15 full names of the authors are missing.

Review #2
By Martin Murillo submitted on 28/May/2013
Suggestion:
Minor Revision
Review Comment:

The paper focuses on an area that has been overlooked/neglected by mainstream researchers and practitioners and provides an important demonstration on how new technology, appropriate technology, and appropriate processes can combine to produce valuable results for subsets of the population that will not be able to enjoy the benefits of “mainstream” approaches in years or even decades to come.

For that, the paper presents a pioneering instance of the appropriate representation of linked data to disadvantaged sectors of the population. The paper is extremely relevant for the special issue and the journal. The paper presents rich technical terms and focuses on issues that are key for this technology to work in the field. The paper tries to compile valuable issues gained in experience and work in the field.

The content of the paper is extremely valuable. However, the presentation needs to be revised to be an appropriate piece of work for the journal and special issue. Please consider making the following changes which I think will make the paper even more valuable and fine contribution in the topic and the area.

1. State whether the paper is a research paper, a report, or other. Depending on the nature of the paper, appropriate conclusions and attention to results, outputs, and/or learning will need to be given. For instance, the paragraph that states “In this paper, we describe our ongoing investigations the implementation of Linked Data-backed solutions for the rural Sahel regions” does not make sense; the term “ongoing investigations” is misleading; the authors need to pay special attention to several instances where phrases such as this might not be appropriate for the context and where the English language is weak.

2. The paper does not seem to follow a methodology or framework for providing and ordered and fluid description of their work. The paper scatters concepts, ideas and processes at times in a disorderly manner. There is also an apparent overloading of technical specifics mixed with subjective issues. One of these instances is the phrase “In this section we describe a prototype demonstrator developed in parallel which exposes the market data gathered through the RadioMarché system using Linked Data approaches, so that new opportunities for product and service innovation in agriculture and other domains can be unleashed” is an example of the scattered ideas and the lack of framework (and weak English). The authors need to focus on topics; I this sentence, for instance, the authors do not need to mention “so that new opportunities for product and service innovation in agriculture and other domains can be unleashed”; this should have been mentioned in an earlier section describing the benefits of the overall approach; they could also repeat in a conclusions or recommendations section. For instance, they could state “The present work illustrated how … new opportunities for … were obtained through the utilization of …”, a clear contribution in the area. The authors need to correct various similar instances present in the paper; doing this the paper will provide important feedback in this multidisciplinary area.

3. I suggest the authors have a separate section that defends their approach and provides with advantages and disadvantages or their choices, an intrinsic part of a paper and scholarly work. This will in turn provide more objectivity, fairness and positive outlook to the paper.

4. The authors repeat ideas; the paper could substantially be cut by avoiding unnecessary repetitions.

5. In a “future work” section, the authors could explore the possibility of sending messages (requests and replies) through text (i.e. SMS); the slightly higher generation mobiles could provide voice through a text-to-voice mobile applications. This becomes relevant in light of the high rates of voice communications in too many nations, i.e. villagers in some rural areas in poor nations pay 0.3 USD per minute while making as low as 100 USD a month. Thus alternatives need to be considered (i.e. a disadvantage of the present approach).

6. “We argue that linking the locally produced and consumed data to (external) Linked Data sources will increase its value” – has a conclusion in this regard been reached? Would be valuable information

7. While the documentation of successes and failures is very important in the area; it is important to make key issues explicit. In this regard, the paper does not discuss future prospects and how to go beyond pilots (a huge problem in the ICT for development area). The introduction of these components would highly make the paper even more relevant as the community does not have first-hand examples on how the introduction of open data and linked data would positively affect the outcome of business models (i.e. the addition of value and/or the use of key component). I understand that this might be beyond the topics, however such information is key for providing better prospects to the “all” to which the issue is focused.

8. The authors mention that “a number of efforts are being undertaken to bridge this so-called digital divide” and the authors mention only one of such instances. If such topic is mentioned and instances identified, I encourage the authors to mention more relevant efforts in that realm. In fairness to other organizations that have sought to decrease the digital divide for decades and because the scholarly nature of the paper, the authors need to do further slight research on the subject, if such in incorporated in the paper. The authors could incorporate instances of international development initiatives (Spider, IDRC, etc), practitioners (nethope, inveneo, etc), and bigger more concerted initiatives such as efforts by the ITU (i.e. current ITU-I WTPF initiative).

9. “In fact, the demand for honey offered became so large that the supply could not cope[13]. For this reason, the RadioMarché system is currently not operational until May 2013” What learnings does this provide to the community? The authors overlook a key problem in these projects, the issue of sustainability. While this might be outside the reach of the paper, this is a great opportunity for providing recommendations; i.e. what would they do different in the future? Is it only an issue of demand for honey? What could “revive” the project?

10. The authors mention “The aggregated data can be used by NGO’s to assess running programs and increase their own transparency and accountability”. This can be put in an earlier sections that describe the benefits. The authors loosely mention “transparency and accountability” without explicitly listing how the technology will increase such.

11. The authors often make use of “multiple languages” expression; in this context, I suggest the adoption of the term “localization” that is widely used in the computer science and in the ICT for development areas.

12. Consider taking the “Why Linked Data?” subsection to an earlier section that describes the fundaments of the platform along its advantages compared to other platforms.

13. Avoid using links in the text; put it as endnotes. I.e. Instead of putting “We use PURLs for the resource URIs. The names-pace chosen ishttp://purl.org/collections/w4ra/radiomarche/” put “We utilize PURLs for the resource URIs with appropriate namespace chosen1” And state the link in such endnote or footnote, as the authors do in other instances

14. “The efforts above represent first efforts into exposing locally produced data in developing regions”. This statement is overly simplistic; I urge to focus on rural poor areas of developing nations. Note that some capitals in developing nations are as well connected as their counterparts in developed nations; thus the statement needs to be more accurate and focused. I suggest adopting the term “poor areas”, “poor nations”, “aspiring nations”, etc.

15. The paper lacks conclusion or recommendations and future work. Current work could be rephrased to “work in progress”

16. I encourage making a revision for the English language.

Naturally the authors can chose not to make some of the changes suggested above. In this case I suggest to state the areas that are out of the scope of the paper; this will let the reader not to have unanswered questions by reading the phrases in the context of the limitations set by the authors. Naturally the authors will also need to be wary of lose and incomplete phrases that will be outside the scope of the paper. Regardless, the authors are encouraged to make a check for the English language.

Review #3
By Michalis Vafopoulos submitted on 01/Jun/2013
Suggestion:
Accept
Review Comment:

The paper entitled “A Dialogue with Linked Data: Voice-based Access to Market Data in the Sahel” introduces a market information system, RadioMarché, as a prototype demonstrator that enables voice-based access to the Linked Market Data. It is an important effort, because engineers practical solutions in bridging the digital divide both in geographical and technological dimension.
Based on the fact that the functional elements of the proposed solution involve economic parameters like prices, offerings, products etc., the authors should consider the case of incorporating the GoodRelations ontology in their model. This could facilitate potential interlinking benefits such as comparisons to international prices and marketing tools.

Typos
Page 2, 2nd column
Our data is initially designed to be produced and consumer by the regional farmers themselves.
“consumed:, instead of “consumer”.

Page 3, 1st column
Space needed before the references.

…system[12].

…could not cope[13].

Page 4, 2nd column
“languages” in …world llanguages.

Page 5, 1st column
Not clear.
Based on the choice, a in the PHP document a SPARQL query is constructed.

Page 7, 1st column
Space needed before the reference.
…in the Bankas area in Mali to Linked Open Data[17].

Also, in Page 3, 1st column, 2nd sentence, the repetition of the full description for the “Market Information System” should be replaced by the “MIS”.


Comments

The revised version addresses the previous comments.

Small details could be still improved though, such as for example:
- pag 6 - "Figure 3 shows a fragment of the VoiceXML code". Check the figure labeling (Fig. 4 on page 6)

- pag 7 - "Example RDF graph snippet of linked market data in the triple store." The reference to this figure is at page 5 ? "In Figure 4.3 (?), we show how in the case of our linked market data audio and literal labels can be mixed, for different languages".