It's not learning fast enough. Can I speed things up?
One way to speed things up is to view an article on the Article Screen, then tap to see its related topics on the Relevances Screen, then set your interests in those topics. Setting topic interest directly will make NewsBrain learn faster.
Another way to teach NewsBrain directly is by going into the Interest Profile itself and setting interest on specific topics. Tap the icon in the top right of the Topics Screen, then pick View As..., then pick List by Name. Now it's easy to scroll through the (couple of thousand) topics, setting the interest levels. Just setting a few of your main interests and disinterests will make a difference. You can also see the topics by Interest Level. Look at the top and bottom of the list and make corrections to ones that seem far off. This will speed up learning. If you view by interest level and change a level, topics may zip around as they change order to stay sorted.
Sometimes I see the same article more than once. Why?
Do you have NewsBrain 1.0 (as seen in Settings->About NewsBrain)? A bug caused some articles to appear multiple times. This is fixed in version 1.0.1. But a few articles will still appear more than once:
This is usually because the source published it more than once. Sometimes a source will make a correction or an update to an article and re-publish it. Sometimes, the change is tiny, such as a spelling correction. But usually you can see a difference, in the published time or the content. If they publish their update as a new article without removing the old, we see it as a near-duplicate.
NewsBrain has subscribed to some feeds that are way off the mark. Can I fix it?
As NewsBrain learns, it keeps track of how "satisfying" each feed has been. If a feed has supplied lots of articles that you find uninteresting, it becomes a candidate for unsubscribing. A feed that becomes very unsatisfying will eventually be unsubscribed. NewsBrain's automatic unsubscribing is purposely a relatively slow process. Auto-subscribing to new feeds is also slow, but it tracks average article interest levels, and will pick up new feeds and drop old ones as necessary to raise the average interest of articles it fetches.
You can make NewsBrain drop a feed directly. A swipe on the Source Feeds Screen will do it. You can also go to the feedly.com website and subscribe and unsubscribe there. Go to feedly.com and login with the same google account you used for NewsBrain. Any subscription changes you make there will be reflected by NewsBrain.
When I page forward and then back again, I often see different articles than were on the previous page before. Why?
The ordering of articles within NewsBrain is always in flux. Most newsreaders fetch and show articles in chronological order, the ones most recently posted to the Internet are shown first. But NewsBrain shows articles in order by how interesting they are, and this changes over time. When you rate articles, this affects the interest prediction of other articles and thus the order they will be shown. And, article predictions take some time, so for a few moments after an article is fetched, it's interest prediction, and thus ordering, will change.
When NewsBrain pages forward, it shows the most interesting unseen articles available at that instant. When it pages back, it shows the articles most recently seen. Thus, when paging back and forth, you usually see different articles.
Am I going to spend the rest of my life teaching this thing?
No. If you like the mix of articles NewsBrain is showing, you don't have to do any more rating or tweaking. Just watch the screen, flick pages with your thumb, and read, read, read. If you want to adjust what it is showing, you can do a little more teaching, whenever you like, as much or as little as you want.
Some Interest Views say minus-zero (-0)! What's up with that?
An Interest View rounds a level to the nearest whole number, so "-0" means the real level is between zero(non-inclusive) and -0.5, where "0" means between zero (inclusive) and 0.5.
Why didn't you include FaceBook or Twitter as topics?
So many articles include phrases like "Follow us on Facebook and Twitter" that those words match a huge number of articles that aren't themselves about Facebook or Twitter. It's the same for some other common words as well.
My city/town/team/interest isn't a topic
There are so many cities, sports teams, celebrities, and so on that they can't all be included, so only a very few are. Add your own favorites! You can add new topics anywhere in the profile's tree. Putting them at the top level keeps them independent, and this is fine. Putting them at lower levels will affect related topics when they are adjusted, so ideally place them somewhere under a topic that is a similar interest with a larger scope.
The whole Interest Profile is in English. Can I get articles in another language?
If you add topics in different languages, those topics, and eventually feeds in those languages, will start to appear. Translating the whole profile into another language should work quite well, but we don't have such profiles available at this time.
I'm very interested in sports or technology or some special subject. Can I get a profile that's more specific?
We don't have specific profiles available, but if you make one, it should work well. For example you could make a sports-only profile that had your favorite sports at top, then leagues under that, then teams under that, and players under that. This approach will work with all sorts of specialties. For example a medical professional might make a profile full of medical terminology, or lawyer might make one full of legal terminology. Such profiles should yield many technical articles on the specialty, eventually subscribing to new article sources based on the specialty. If you make a good specialty profile, it can be exported and shared with others. You can have several profiles, and activate the one of interest. Incoming articles are evaluated against the active profile.
Some articles say No Relevant Topics when there should be. Why?
There are generally two reasons this may happen. If the headline and the article are very short, it might actually be that none of the two thousand topics in the Interest Profile match. But it is also possible that NewsBrain hasn't yet completed its Topic Matching process for that article. Every article is scanned thousands of times to check its relevance against every topic in the profile, and depending on the number of articles waiting to be seen (typically one hundred or more) and the model of device you are running, it could take anywhere from ninety seconds to five minutes to fully complete all the perhaps hundreds of thousands of scans. NewsBrain fetches articles ahead of time to give it enough time to complete its prediction before you get to it, but this varies. Basically, the article will get new relevant topics attached soon, often even as you are viewing it. The more relevant topics attached to an article, the more your feedback on the article will affect learning and future predictions. But even an article with no relevances may be rated - the general satisfaction level of the feed it was fetched from will be affected, and that affects the level of all the articles it fetches.
The screen says No Articles. I tapped the Refresh button but no articles came. Why?
If the refresh button has no effect, it probably means that all of the articles posted by all of the article sources have already been seen. Is it the weekend? Significantly fewer articles are posted on weekends. NewsBrain typically subscribes to 40 or more sources, and a new batch of articles will generally arrive in a matter of seconds. NewsBrain will notice the dearth of articles and may eventually subscribe to more sources to keep your magazine endless.
Does the app use a lot of energy or bandwidth?
NewsBrain does get quite busy at times. While you are browsing articles, it is busy predicting interests, fetching more articles, and so on. So, the CPU is busy, but when you close it or switch to another app, it pauses. It doesn't do any work in the background while it is not active, so it isn't using up your CPU or battery at that time.
The app fetches the summary text of about 100 articles at a time, and when you have browsed through most of them, it gets another batch. When an article comes on the screen, the picture for the article, if any, is fetched at that time. Again, the app only communicates with the Internet while active. When you aren't using NewsBrain, it uses no CPU or bandwidth.