Search Mailing List Archives
[bioontology-support] NCBO Recommender
Marcos Martinez Romero
marcosmr at stanford.edu
Thu Jan 12 17:10:29 PST 2017
Hi, Clement, Amina,
Clement’s solution would be a quick and easy way of aggregating the results from multiple calls. For some situations (e.g., the same term appears in different input fragments, a multi-word term is splitted and put into different input fragments), the final score obtained will be slightly different from the score obtained by one single call. However, without doing a deep analysis on this, I think that the ranking of ontologies obtained shouldn't be significantly different from the ranking obtained using just one call to the service.
On Jan 11, 2017, at 2:47 PM, Clement Jonquet <jonquet at lirmm.fr<mailto:jonquet at lirmm.fr>> wrote:
Indeed it seems to be a to big piece of text. Can you split that in several ones ?
Marcos, what would be the best way to aggregate the multiple results from the recommender into a unique one ? Is a basic average (mean) of each ontology score a good way ?
Any other advance approach ?
De : Amina ANNANE [mailto:a_annane at esi.dz]
Envoyé : vendredi 6 janvier 2017 01:20
À : Clement Jonquet <jonquet at lirmm.fr<mailto:jonquet at lirmm.fr>>
Objet : NCBO Recommender
I tried to use the NCBO Recommender through the REST API for the text in the file attached to this e-mail. Unfortunately, I had no answer from the server. I do not know if the problem is related to my internet connection, or to the text (a lot of words...).
I want to use the NCBO Recommender to pre-select a set of ontologies. From this set, I will apply a strategy to build a BK. However, if the NCBO Recommender is limited I have to think about another solution
<Text for the recommender.txt>
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the bioontology-support