Sources and Queries
No matter the subject or scope of your watch, you will need to set up a monitoring system that will collect information for you. Digimind monitoring agents are a combination of the sources you would like to track and the query you can add to filter out information you are less interested in.
For a monitoring agent to work well, there needs to be a balance between the sources and the query. For very general sources covering a wide-range of subjects (for example Twitter or query sources) you will need to prepare a filter to protect you from the information that is not about your subject. For very specific sources that cover a more focused subject (for example a specific company’s website or a specific RSS from a specialised site) your filter will be much less specific, and in some cases, you may not even need to have a filter.
When you first set up a monitoring, you will have a fairly good idea about the keywords you need to include in the ftiler in order to balance the sources you track. Nonetheless you may find that you get some surprises and that you need to fine tune.
There are two types of problem:
- Slience: the monitoring is not bringing back enough information or the information you need
- Noise: the monitoring is collecting information that is not useful for you
In both cases, you can apply the same methodology.
1: Wait and observe
Wait a week and see what you get back over a full working week. The reason for waiting is that some of the RSS feeds or other sources in your agent may be updated once a week, and you will need some material to work on.
2: Make choices about the information
During this week, continue to save any of the information that interests you into your folders and projects.
Do not delete any of the alerts for information that is out of your scope. You will need this for the next step.
3: Analyse your choices
At the end of the week, look back at the information you have decided to keep.
Silence can be the result of either the sources or the filter in the monitoring. Look back at the information you have validated or stored into the platform. Keep an eye out for keywords or set phrases that are in these articles but are not yet in the filter. If needs be you can add these words and phrases to the filter. This will help you to pick up more information by extending the reach or power of the filter. For sources, check you have covered everything. Add some more sources to the monitoring agent.
Noise is often generated by the filter. Take another look at the information you decided not to keep. Try to work out how this piece of information got through the filter you set up. Keep in mind that you have received the information because you asked for it. One common cause of noise is that the keywords in the filter are not used in the right context. Add in keywords to contextualise:
- Orange = a color, a fruit, a town in the south of France, a telecommunications company
- Orange AND (telephone OR telecommunications OR communications) = a telecommunications company
- Batman = a superhero, a town in Turkey
- Batman AND "DC comics" = a superhero
- Batman AND "DC comics" AND (book OR ebook OR comic) = only comic books about my favorite superhero
Another good technique is to position your keywords, either in the article or close to each other. The intitle: and begin: operator will position words in the artile and the tilde ~ will place the words close to each other.
- intitle:batman = only stories with my superhero in the title
- begin:batman = I will receive the article if Batman is in the first 25% of the article
- intitle:batman AND begin:"DC comics" = Batman in the title, contextual words in the first 25%
- intitle:(batman OR "Dark Knight") AND begin:"DC comics"
- Batman AND film = I will receive the information if both words are somewhere in the article, one at the top, one at the end, anywhere will do.
- "Batman film" = The query is asking for a fixed expression, both words in that order.
I have gone from one extreme (anywhere in the article) to another (a set phrase). These choices can be very useful but at times I might need to put the words close together, but not necessarily in a fix expression. This is when I can use the tilde.
- "batman film dc"~10 = I need all three words inside a group of 10 words.
On occassions you will also identify sources that are generated a lot of noise or unwanted information. Here you have two possible actions. Firstly, you could decide to remove this source completely and no longer track changes on it. Alternatively, you might want to keep the source, but move it to another monitoring agent for which you will create a different filter better suited to the information generated by this source.
4: And now you wait again.
After you have modified your monitoring, wait another full working week and then start the process again.
Depending on your needs and the sources you have decided to track you may need to repeat the process three times.