Like a lot of DX chasers, I spend far too much time watching DX clusters, hoping that the next spot might be something I actually need. The trouble is that once you've worked a couple of hundred DXCC entities, almost everything scrolling past is for somewhere you've already worked. You still keep watching because every now and then something interesting pops up, but there's a lot of noise in between.
That got me wondering whether there was a better way.
The idea was pretty simple. Could I build something that looked at my logbook, watched the live cluster feeds for me, and only showed stations that might actually move me forward?
 |
| Version 1.0 of my personal DX spot monitor |
I'd been curious about using AI for software projects, so this seemed like a good excuse to find out what it could really do. Rather than asking it to write a program and hoping for the best, I treated it more like another developer sitting alongside me. We'd build a bit, test it, break it, fix it and then think of something better.
The first version is now working surprisingly well.
It runs locally on my laptop and connects to both the UK DXSpider cluster and the Reverse Beacon Network. I simply load an ADIF export from my log and, from that point on, it watches the incoming spots and compares them with the DXCC entities I've already worked.
If a station belongs to an entity I've already logged, it quietly ignores it. If it's something I still need, it appears on the screen.
That sounds like quite a small thing, but in practice it makes a huge difference. Instead of watching hundreds of spots drift past, I'm left with a much shorter list that's actually worth paying attention to.
So far the application can:
- Monitor DXSpider and RBN continuously.
- Reconnect automatically if either feed drops.
- Read a standard ADIF log.
- Resolve spotted callsigns to DXCC entities.
- Filter out entities I've already worked.
- Filter by band, mode, source and spot age.
- Remove duplicate spots automatically.
One thing I particularly like is that everything runs locally. My logbook never leaves my computer, there are no subscriptions, and it barely seems to make the laptop notice it's there.
Of course, as soon as the first version started working, the list of ideas got much longer.
The obvious next step is to make it less like a filtered DX cluster and more like a personal DX assistant.
Some of the things I'm already thinking about are recognising major DXpeditions automatically, highlighting genuinely rare opportunities, keeping track of countries where I've worked the station but still need a QSL, adding desktop notifications for the stations that really matter, and perhaps even bringing propagation data into the picture so the software has some idea whether a station is likely to be workable from my location.
I'm also starting to think about longer-term features like keeping a history of every spot received, tracking award progress beyond basic DXCC, and generally making the software a little smarter about what deserves my attention.
Perhaps the most interesting part of the project hasn't been the software itself, though. It's been discovering what AI is actually good at.
It turns out it's much less about pressing a button and getting a finished program than having a conversation with something that never gets tired of refining an idea. Every time I think "it would be nice if it could also..." we end up exploring another approach, trying it out and seeing whether it genuinely improves the application.
It's still very much a work in progress, but it's already become something I leave running while I'm on the radio. That probably says more about whether the idea has worked than anything else.
No comments:
Post a Comment