So I have gotten a lot of feedback and learned a lot during the past few days. I am confident that by utilizing open source methods and collaboration there are possibilities to find solutions and or make fascinating experiments concerning:
- text input in VR
- mitigating RSI
- speech recognition
- improving posture
- international input
By creating a framework for experimenting with various technologies and approaches. However, for me to continue working on this I must call on your support. Please consider donating:
If you like what I am doing but can’t or won’t support me via Patreon sharing what I am doing would help me reaching out to more like minded people who can come with suggestions and ideas!
I am currently staying in Turkey and am traveling the world for two reasons. One, to keep costs down and two, in order to get hands-on experience with different languages and the quirks that come with them. My eyes where opened to this while living in Greece, and it would be a shame if ancient writing systems where to die out due to them not fitting in into an American worldview. Even German and Swedish is affected so it’s even an Eurocentric worldview.
In other news
Apart from that I’ve been working on a prediction engine under the name of Odin. At the moment it is basically a good enough solution that I successfully used for earlier attempts.
In short, it is similar to a PPM prediction system. So it keeps track how many times a string of up to n letters lead to a specific symbol. Given a string it can provide a list of the most probable letters – or rather all of them, sorted by frequency.
The code is written using Go and I’ve set up a rudimentary make system to build frequency profiles in json format for later usage.
Perhaps the most interesting idea here is to make use of separate dictionaries. These can then be weighted and summed up. The weights would be changed after dictionary hits, making a switch to a different dictionary as simple as starting to write content from it.
I’m toying with the thought of adding more data to the wordspace such as time of creation, geographical location and of course work from where it originates. That would mean that it would give better predictions if it figured out that you are writing in an old fashioned way for instance.
So in the coming days I’ll try to compose a summary of the brilliant ideas I’ve gotten from readers as well as to work on some HTML5 demos that you can try easily in your browser. They all depend on having a prediction engine so know when that is in place it won’t be long.
Don’t hesitate to comment, even if you think what I am doing is pointless and stupid I’d rather hear your arguments than live in ignorance!