A few brave souls have been giving the first version of QSObot some exercise, and, looking at the logs, I could see that some basic things were too fussy or just broken, so I’ve made some improvements.
The recognition of cq calls is a bit easier, now; before it was demanding perfection, now it can pull apart words like “cqde” into “cq de” and proceed with the parsing.
I’m also checking callsigns, names, RST reports, and QTHs for sanity, so the bot doesn’t accept crazy versions, or overwrite good versions due to keying trouble.
To ask for the bot ‘name’, some people were asking “yr name?”, which I expected, but some were asking “yr name ?” (separate ‘?’) and even “yr name pse ?” (‘?’ is a long way away from the word ‘name’). The bot can now decode this sort of thing a bit better, so should be less frustrating.
It can also understand “ur 599” as well as the more formal “rst 599” or “yr rst is 599”.
At the moment, the bots are all about getting the basics exchanged (call, RST, name, QTH). Once this is done in both directions, they go into a ‘rag-chew’ mode which I haven’t done much with, yet :-)
In rag-chew mode, the bots just send random Mark Twain quotes at you, until you give up and close the QSO.
Next steps will be to:
- add more questions and answers to each bot, like rig, ant, wx, age, job.
- make the rag-chew mode more interactive (this will be hard, I think!).
Please give it a go, if you get the chance. Any feedback to me, through the Add-ons group, or directly, would be gratefully received.
Slightly updated instructions are on the Using QSObot page.