Highlighting some new visualization functions in Lens 6.2.0 — Three use cases for Lens.org
I have become quite a fan of the free Lens.org since last year , writing pieces like - 7 reasons why you should try Lens.org that talked about the features in Lens that make it worth a try.
Since then there has been quite a few major updates including the release of an API that pushed Lens even further ahead. In this piece, I’m going to share with you 3 real life scenarios where I have used Lens.org for analysis.
Here’s the scenario. So there has been quite a bit of ruckus in the Scholarly Communication world due to Plan S and the transformative agreements pushed by various Consortiums including UC (University of California).
This has created waves all around the world including here in Singapore and this had led us to all sorts of questions ...
While one can use the usual Scopus or Web of Science to try to answer these questions, I have started to favour Lens.org more due to the amazing visualization capabilities built into it , together with the ease of saving and sharing of dashboard.
This post will cover 3 use cases I have encountered and highlight 3 functions that was useful for me to answer the question namely
- Use of custom groups
- Use of “restricted filtered values” function
- Reuse of custom groups , particularly in visualization types that don’t normally allow custom groups e.g. grouped bar chart, stacked bar chart
1. What proportion of Singapore’s output is published in various publishers in 2018?
Transformative agreements like Read and Publish or Publish and Read depend on the existence and knowledge of various data. In particular, the expected amount of publications published annually in each publisher is among one of the most important factors.
Take Publish and Read agreements between consortiums and publishers, roughly speaking the idea is that the consortium will take the total subscription fees normally paid by the institutions to publishers and instead use it to pay APC (Article processing charges) for each paper published by their authors so that it can be made open access immediately.
So for example, if a consortium normally paid $5,000,000 a year for subsciptions to say Wiley and has authors who publish 5,000 articles annually, the APC may be set at a baseline of 5,000,000/5000 = $1,000 per article. Of course in actual fact there will be tons of negotiations as some publishers will not be happy with this and the final deal struck will often be at a higher price point, because insitutions will also be able to read all Wiley content on top of making their papers Open access, which publishers argue is worth something.
But no matter the final price, you can see why the number of papers published is a key point to figure out.
So one of the questions I had was if Singapore wants to do a Publish & Read deal at the national level (say like the Publish and Read deal Norway -which has roughly the same annual output as Singapore with Wiley) what would the target APC price be like?
We libraries of course have the subscription costs but this hypothetical consortium will need to figure out the numbers of papers published for each publisher in 2018.
Lens makes it a breeze of figure it out , visualize and share it.
Firstly, I restricted the results to only publication with Institution Singapore and for 2018 only. You can use Lens structured scholar search (essentially a advanced search) with a interface that any librarian would be very familar with or you can just use the filters on the left hand panel of Lens to slice and dice.
For papers with Singapore affiliated institution authors and publishing in 2018, I get around 18,000 records. So I click on the analysis tab to search some visualization
There are some preset visualization options that create visualizations for Authors, institutions, funders, journals, but I don’t see anything for publishers, let me click on custom instead which gives me 11 types of visualization.
Let’s try something simple and select “Horizontal Bar chart”
By default it gives me Top institutions by document count, so let’s change it to “Publishers by document count” and this is what I get
It looks pretty nice but then I notice, the data labels aren’t that clean. For example,you can see rows for “Elsevier BV”, and “Elsevier” “Elsevier limited” in the top 20 publishers, which obviously should be combined.
The great thing about the latest version of Lens is you can now group these stray labels together. First click on setting on the top right to reveal more options.
To group them click on Custom Groups and the rest should be intitutive.
Hint : By default you see only the top 20 items. For greater accuracy it might be a good idea to increase the new limit to 50, 100 or 1,000 before you start doing custom groups and then changing it back to the top 20 (if this is what you wanted).
Once you are satisfied with the groups, click “Apply Groupings”, add a title and you get a very nice graph with your grouped values. You can tell it is grouped by the yellow triangle icon next to the title. You can click on it to see what custom groups were made.
Once you are happy with everything you can save the graph as a new dashboard and share it publicly (or not).
You can edit your saved dashboards under Work Area -> Analysis Dashboards
2. Finding publication share of papers per institution in Singapore
So say this hypothethical Singapore national wide consortium decides to take a united front against publishers and say a deal is signed. However internally there will be a need to allocate costs.
Assuming a Publish & Read type of deal, cost allocation will logically not be based on FTE (full time equalvants) but on publication output. So let’s use Lens to visualize this.
As before, filter results to only Singapore Publications in 2018. To keep things simple I’m going to use the exact same visualization type. But when I run it, I notice something odd. While the expected Singapore institutions are there , I see non-Singapore ones like “Chinese Academy of Science”, “University of Cambridge” etc in the top 20.
You might think the solution is simply to filter first (as in putting a checkbox next to Singapore institutions in the search) by only Singapore Institutions and then visualize but you will still see the same result.
The reason is because even with the filter, you will still be getting papers where the co-authors are not all from Singapore affiliated institutions and this shows up when you visualize. Trying to filter out non-Singapore institutions in the search (as in putting a cross against non-Singapore institutions in the search) is not the answer either as you will lose papers that you want in the set.
So what’s the workaround? Answer: once you have filtered in the search to only Singapore institutions (as above), visualize as per usual, click on settings (on top right) of chart and then check “Restricted to Filtered value”
This will get you exactly what you need.
To recap the steps are
- Restrict your results in the search to only Singapore institutions
- Visualize and click on settings
- Click “Restricted to Filtered value”
3. Number of papers acknowledging major funders in Singapore, segmented by Field of study
Funders who are thinking of following the Plan S mandate of making research funded by them to be made immediately open access via Open Access Journals or Open repositories may also find Lens useful to ascertain statistics like how many papers this will affect, or % of papers that are already open access.
While doing this with Lens, the main thing you will notice is first the data for Funders in Lens is pretty dirty, so you will need to do a lot of grouping. You may sometimes wish to reuse the grouping for other charts, but currently you can’t save groupings directly, but I will show a workaround to this.
Secondly as you explore the visualization for Lens, you will note , currently customized groups only works for some visualizations. For example, you can’t use custom grouping for stacked bar charts or grouped bar charts. Or can you? Let me show you a workaround.
Again we will be working on publications by Singapore affiliated authors who have published in 2018.
In this example, let’s try to
- Create a simple bar chart showing —Papers by Singapore Funders (Custom group)
2. We will then reuse the custom grouping in a grouped chart.
But let’s talk about #1 first. Similar to the example with papers by institutions, we want to only show Singapore Funders and not other funders , so let’s filter by funders first before visualization
Filtering for researcher funders is tricky
Filtering for funders is pretty complicated because there are quite a few of them and the labels are not really controlled.
I recommend clicking on the “More” button, and this is what you will get.
The first obvious thing to do is to scroll down the results and select the ones you want and click refine. However, you will be picking only from the top 100 labels which is often enough. But what if it is not and you want more accuracy?
However you can try to improve this by using the search option to try common variants of funders you want for more accuracy on.
For example in Singapore the major researcher funders are Ministry of education often abbreviated as MOE and National research funder NRF. This can be quite mundane work , depending on the accuracy you want. For full accuracy you may even want to search with different order of words if there are so many matches it doesnt list everything (hint if your last option shown is say (15) chances are there are entries that match that are not showing. )
But once you are ready, you can start visualizing and grouping as per in the first business case.
It goes without saying the data is extremely messy, due to the lack of the use of a standard identifer (Fundref) but I did the best I could
Wow that’s a lot of working doing the custom groups! With all the work done doing the grouping, how do you reuse it?
There is no obvious way to do it right now, but here’s a trick that Lens Staff shared with me.
What you need to do is go to settings ->View source->Lens Chart Config and copy the JSON, in particular the part from grouping onwards.
With this saved, you can then create any brand new chart and reuse the custom groupings by pasting back the JSON.
This even works for graphs that do not normally allow groupings!
To show this, let’s try to create a grouped chart, funding by field of study with the same set of results.
You can quickly create one , but by default you can’t directly create custom group.
So once you have that, you can add back the custom funder groups we copied earlier. Go to settings -> view source ->Lens Chart Config and add the json part from “groupings”: { — we copied earlier on
Then Click “Import from Config” and you will get the same graph as before but using your custom funder groups (notice the yellow triangle icon next to the title appearing).
To see the saved dashboard .
Challenge : Notice that it lists funders that are not from Singapore , can you fix it?
Conclusion
I’ve just scratched the surface of what Lens.org is capable of. Hopefully the 3 use cases I have featured and the functions I have shown will inspire you to do your own visualizations.