search

theareaorg / AREA-Research-Agenda

Creative Commons Zero v1.0 Universal
3 stars 2 forks source link

measure of risk-reward for each reach topic #16

Open wzbernstein opened 3 years ago

wzbernstein commented 3 years ago

use cosine similarity measure to compute how similar descriptions are compared wiht abstracts of databases. The average (or summation) similarity of each description to all abstracts can serve as some measure of risk.

wzbernstein commented 3 years ago

Initial Results of Evaluating the Amount of Previous Work Per Research Topic

Low Power DSPs for Use in AR Display Devices | 5.0 Common APIs for Tracking Libraries for Vision-based AR | 4.1 Visualization of Temperature Sensor Data for Emergency Response | 2.8 Impact of Spatial Vision on Visual Encoding and Memory Anchoring | 2.6 Off-Device Fusion of Streaming Data from Diverse Sensors | 2.3 Informing Users About Hazards in Proximity | 2.2 Documenting Water Pipeline Network Topology via HMDs | 1.7 Operational Risk Categorization/Matrices as an Indicator of AR Impact Potential/ROI OR AR as a Mechanism to Effectively Mitigate Operational Risk | 1.2 New Employment and Crew Models in Metals and Mining Value Chain | 0.1

wzbernstein commented 3 years ago

@j-derby @cperey @peterorban @bchap-erau | Please comment here. Do you think the rankings above make sense?

theareaorg commented 3 years ago

Thanks, Bill! I'm still unclear on what "risk-reward" means. I imagined a two by two system with risk on one axis and reward/impact on the other. Similarity with past research = risk?

Leaving aside for the moment what it's called, we're clear that this process compares the research topic descriptions with all (or some?) of the abstracts in the database for similarities.

I must say that I am somewhat surprised that there is 100% overlap (highest is 5 correct?) for the first topic. I'm also very surprised that there are so many past research publications about common APIs for tracking libraries. Perhaps there are MANY papers about tracking libraries for vision-based AR! That is my impression and that is EXACTLY the problem/issue that this research topic would try to addrss.

I'm not surprised that the lowest two are low. These are important topics that the scientific community has not looked at and is unlikely to ever examine...