bias in shape recognition-description

Pns27

New member
Joined
Mar 24, 2020
Messages
2
Hi there,

I am Panos, I am a PhD candidate in Archaeology, so as you can guess maths is not the field of my expertise. Therefore, please forgive the ensuing ignorance on some math's issues.
Before my question, I should give you a little bit of context. I study Neolithic-EBA boats constituted by 2D and 3D representations (there are no actual boats/shipwrecks) of the area I am studying. It is a joint programme between Archaeology and Ship Science, so, the main goal is to classify them, identify the boat type they represent and reconstruct-test their properties with Naval Architect software (MAXSURF).
So, my issue is the bias on the description of the data. Before any analysis, I have to do a descriptive analysis based on the morphology of the boat evidence, i.e. I will deconstruct them on their basic, discernable components (bottom, posts, gunwale, sections, etc.) which I will eventually code (to run a multiple correspondence analysis.
However, how do I describe a shape without being biased towards it? I mean, for me, a sten-post might be constituted by an asymmetrical, oblique line, but for someone else not.
Does anyone have any insight on the matter? Is there any software identifying the shape based on mathematic formulas? Any other solution/workaround would be much appreciated.

I hope it does make sense.

Panos
 

afos

New member
Joined
Mar 3, 2020
Messages
23
Such problems are typically addressed using the method known as a committee of experts. 1 person may be biased. 7 people are likely to be less biased as a group. Get 4-10 more colleagues of yours to describe the boats and then, for each boat part, choose the median description. If the descriptions are quantifiable (can be turned into numbers), you may experiment with using the average of opinions as the overall opinion of the committee.
 

Pns27

New member
Joined
Mar 24, 2020
Messages
2
Such problems are typically addressed using the method known as a committee of experts. 1 person may be biased. 7 people are likely to be less biased as a group. Get 4-10 more colleagues of yours to describe the boats and then, for each boat part, choose the median description. If the descriptions are quantifiable (can be turned into numbers), you may experiment with using the average of opinions as the overall opinion of the committee.
Well, that's a splendid idea, especially if include people of various disciplines. Thank you, Afos, I really appreciate it.
 
Top