I’ve purchased the DEVONagent Pro at around half year ago and used it much. I want to write my personal view about this software here, to ask for any suggestions and hear about anybody who has the same view with me.
Well after hundreds of trying and tinkering with DAP, I’m quite tired with its page-ranking AI. For 70% or higher chance it won’t give me the best results for searching.
To give an example, I want to know something about my favorite mac notebook app “Curio” so I launched the search “curio AND app”, with plugin “Google”, at the same time I searched google for the same query. The results are shown below.
Clear to see that Google (.com) knows exactly what I want so it returns the index of zengobi. Well let’s check what DAP returns. At the first place it is a government website and the second one is somehow a traveling website. Strange that they both have the highest ranking scores of relevance.
About each time I used DAP for searching, I got the results just like this case, that is, not relevance to what I want.
In this case I highly doubt about the ability if DAP’s AI. I don’t quite know about how it is implemented but I hold the same feeling with this one: https://dellu.wordpress.com/tag/devonagent-vs-google/
Now Google holds millions of TBs of data and they’ve managed to cast the newest AI technologies (big data algorithms, deep learning algorithms) to rank their pages. So how can a ~25MB app to give better performance than it? In the 25MB app it won’t be anything like semantic database or large corpus so any knowledges is extracted from the search result, it’s much a small corpus for machine learning.
Sometimes I think Google returns the best page ranking result and DEVONagent ruins them all. I say “Hey DEVONagent, why not leave the result as it is ?!”.
I may be expecting too much about DAP, after all it’s built by a small company and we cannot expect it to fight with Google.
The second thing I’m not satisfied with DAP is the language. In the website of DEVON technologies http://www.devontechnologies.com/technology.html it said one of the features of DEVONtechnology is, it’s language-independent.
It is NOT, as far as I know.
There are many types of languages in the world. English is (naturally) split by words while Chinese and Japanese does not. In DAP, it seems, words are sent to AI to perform relevance calculations. So DAP becomes useless when used to search for Chinese or Japanese, or some other asian languages, since DAP cannot judge the relevance of two sentences in Chinese.
There may be misunderstandings or misusage. Any suggestions are welcome here.