Automated text creation
Posted: 18 Apr 2018 12:39Just found a fun website to play with: botnik.org
This uses small corpora from specific genres to create word lists. If you want to write something within the same genre, you simply choose words that appear from the corpus. You can also upload your own corpus. It's easier to play with than explain.
Anyway, I decided to start with something straightforward. I have a textfile of the university slogans of all Thai universities, so I uploaded this to try to create new slogans. Here are 3 new university slogans (created mostly by choosing the first choice of word given):
Quality and morality contributed to local development
University of learning for community
Education is growth of the land
This seems to work, so I thought I'd try something a bit different. Given that so many of you seem to find contributing to this forum difficult, I copied and pasted the contributions from the recent discussion on 'Research and forensic investigations'. Taking only the first choice of word given, the program produced: "The data collection and analysis stages of the research process is similar to a detective trying to solve a crime" (slightly ungrammatical but clearly relevant and a neat summary of the discussion). So it now appears that, even if you have no idea what to write, you have no reason to not make a contribution to the forum.
My final plan concerns the ridiculous pressure to publish in academia. Being in the fortunate position of having already published numerous articles, I realised that I didn't actually need to do any research or write anything myself to create a new article. Rather I could simply upload previous articles and let the program generate a new article for me. Publishing made easy. So I uploaded my 5 most recent abstracts (all on different topics) to produce a new abstract:
The research topics with the lowest proportions of articles arguing that multidisciplinary texts can provide guidance are rarely practical. However curiosity or desire for deep understanding is dependent on strategic competence. This is still relatively rare in Thailand. Alternatively such findings can exacerbate inequalities. Furthermore education policy involves using concepts from different disciplines.
I hope you agree that this is the most valuable and meaningful work I have ever published.
Why don't you try creating something?
This uses small corpora from specific genres to create word lists. If you want to write something within the same genre, you simply choose words that appear from the corpus. You can also upload your own corpus. It's easier to play with than explain.
Anyway, I decided to start with something straightforward. I have a textfile of the university slogans of all Thai universities, so I uploaded this to try to create new slogans. Here are 3 new university slogans (created mostly by choosing the first choice of word given):
Quality and morality contributed to local development
University of learning for community
Education is growth of the land
This seems to work, so I thought I'd try something a bit different. Given that so many of you seem to find contributing to this forum difficult, I copied and pasted the contributions from the recent discussion on 'Research and forensic investigations'. Taking only the first choice of word given, the program produced: "The data collection and analysis stages of the research process is similar to a detective trying to solve a crime" (slightly ungrammatical but clearly relevant and a neat summary of the discussion). So it now appears that, even if you have no idea what to write, you have no reason to not make a contribution to the forum.
My final plan concerns the ridiculous pressure to publish in academia. Being in the fortunate position of having already published numerous articles, I realised that I didn't actually need to do any research or write anything myself to create a new article. Rather I could simply upload previous articles and let the program generate a new article for me. Publishing made easy. So I uploaded my 5 most recent abstracts (all on different topics) to produce a new abstract:
The research topics with the lowest proportions of articles arguing that multidisciplinary texts can provide guidance are rarely practical. However curiosity or desire for deep understanding is dependent on strategic competence. This is still relatively rare in Thailand. Alternatively such findings can exacerbate inequalities. Furthermore education policy involves using concepts from different disciplines.
I hope you agree that this is the most valuable and meaningful work I have ever published.
Why don't you try creating something?