I believe that this research may be useful to you.

• Keywords analysis on Dengue disease from 2005 - 2014: A scientometric study

Abstract

The study has examined the research output of the keyword analysis of 'Dengue disease' during the period 2005 - 2014. Bibliometric indicators were existing journal articles, cited reference, authorship pattern, document wise distribution, journal self - citation and author self-citation, keyword analysis in the dengue disease. The keyword analysis of Dengue Disease for ten-years, during the year 2005 to 2014 taken for the study 3517 words were retrieved in the current study. The highest publication is 691 in 2014 rank is first, the second rank is 2012 in 518 records, third rank is 2011 in 414 records and lowest record is 131 in 2005. The out of 3517 articles single author contributed only 216 (6.14 %) articles while the rest 3301 (93.86 %) articles were contributed by Multi authors. The majority of authorship patterns are multi-author collaborations which is the highest contribution during the ten years. The degree of collaboration ranges from 0.90 to 0.97 and the average degree of collaboration is 0.94. Researchers identified the total number of Countries i.e. 130 during the period between 2005- 2014. But, we observed the majority of the Country "USA" in 1219 records with TLCS- 6211 and TGCS in 27146. Researchers identified the total number of keywords i.e. 5493 during the period between 2005- 2014. But selected top 20 keywords for the research purpose. We observed the majority of the keywords "Dengue" in 2038 records with TLCS 10794 and TGCS in 33761. Moreover the present study mirrors the actual published results of the work of Dengue Disease in global level.

Dear Debrayan Bravo Hidalgo

Thank you for your response.

I have analysed the influential aspects such as top author (their affiliated country and institution), top journal and top documents.

I also analysed intellectual structure and conceptual structure of the select research area.

However, my question was about some systematic and objective methodology to identify the future research agenda. Your referred article just explored the influential aspects that is not my focus of point.

Thank you.

You should use the first quadrant of the thematic map to explore the emerging or dropping topics. this can serve your future agenda. or you can use thematic evolution and provide a recent cut off point to analysis how the underlying topic developed over time this will help to understand the future agenda of a specific field.

Thank you Adeel Nasir

I used this approach and identified the keywords in first quadrant of thematic map. However, those keywords not providing some concrete idea that can be the future research agenda.

What are your views about the dendogram to identify the long distance keywords and suggest their combination as future research agenda.

In other words theory A + theory B = theory C.

Predicting the future by the past is logically impossible. What people do is hoping that the recent past is a predictor for the near future. Yet, one can never predict the really original future contributions.

So, looking at policy documents is a better option. They tell you the problems and questions for which answers are urgently needed (and for which funding is available).

For keyword analysis and future predicting can use Natural Language Processing in Python, R Software. But not possible to say exact the reality. Umar Farooq Attari sir

I think You can use Burst number in CiteSpace to find keywords with high Burst values and consider which keywords might be future research directions in combination with their time variations .

Elaborating on Rousseau's answer- you can predict exceptional growth for a known topic (maybe 3 years out) but you can't predict truly original future contributions unless you pick up the early years (and make an extrapolation from there). Burstiness hasn't worked. There was a huge effort by IARPA (the FUSE program- Foresight Using Scientific Exposition) that tried to model this in terms of Burstiness- with no tangible success. I like Rousseau's suggestion (discussion about new funding initatives). I've been intrigued by the possibility of looking at signals from research proposals (especially the proposals that aren't funded). Another approach is citance analysis (Henry Small's work- building on Teufel).

We have done this a couple of times by building a correlation plot among keywords' and 3-word terms co-occurrences. Keywords that have not previously been mentioned together could mean that some of the topics associated with those keywords have not been concurrently studied. So keywords with both no previous correlation (or statistically non-significant correlations) and an ontological association could provide ideas for future research opportunities. However, you need some knowledge about the field to identify whether the keywords have an ontological association or not. Clearly, this does not predict what is going to be relevant in the future but it's a starting point for a discussion on future opportunities.

Rodrigo- I've seen this in the form of link prediction- which may be a good idea for possible avenues to explore. But I haven't seen any rigorous test of link predictions (such as Threat Score or CSI Score- which is used to predict unusually weather events- like storms). My concern is that these methods suggest what new links might occur- and aren't evaluated by the number of correct vs. incorrect predicted occurrences. Useful for idea generation. Not useful for predicting where emergence is likely to happen. What's your thoughts on this?

Hi Richard. Before your post I've never heard about link prediction. It certainly seems like a great option to explore topics for future research but I would need to read more about that to give any informed opinion. Thanks for sharing your ideas!

I find keywords to be a pretty poor guide to knowing what is in a paper, especially in terms of its original contribution to knowledge But then, I always have great difficulty deciding what keywords to use for my own work!

Then, there is the question of identifying gaps - looking for spaces between keywords and absences of association of keywords may indicate something like "structural holes", but there are often going to be wide topics that have not been investigated - some of which should and could be, but are not because academic incentives push people towards following earlier studies' patterns. Of course, sometimes we simply lack empirical evidence, measurement instruments, or theoretical apparatus at the present time, so topics will have to remain unexplored for the time being. But in areas I work in, I am often amazed by the ruts that research gets stuck into, and the vital topics that are not addressed - and could be!

I am 100% in agreement with the answer of the researcher Ronald Rousseau .

Agree with Ronald Rousseau but in my opinion, a combination of both analysis of the past and policy trends is the best way. Path dependence may affect in some way the policy feasibility as well.