Fine tuning the filter and the sources in the monitorings you create to track subjects on the internet is an important part of the monitoring process.
These tips and techniques will help you do this with more precision and speed.
1. Check for the obvious: the filter
It’s obvious, but a common source of noiset can be as simple as a spelling error in a keyword. Inside a Digimind monitoring agent, I can click to active the spell check:
I can see underlined in red that I have a spelling mistake in this filter. Also check for words that are correctly spelt, but in the wrong context. In this example “baking” is well-spelt, but is out of context as I need to focus on “banking”.
2. Check for the obvious: sources and needs
This monitoring agent has been created to track what the national press is saying about the big banks in the UK. Although a good source of information about these banks would be Twitter or their own corporate sites, neither of these sources are the best way for me to find out what is written in the national press.
In addition, the filters I would need to apply to banks' own corporate newsrooms and to Twitter are totally different. It is worth keeping different types of source in different monitorings, as this will help you to create a filter that works for each type of source.
3. Check for coherence between sources and filters
When the national press talks about banks they refer to the banks by their official names. However, these banks own corporate sites often publish material about what they’ve been doing, without using their own name; instead they use pronouns such as “we”, “our” and “us”. If I were to include the search term “Lloyds TSB” in an agent that monitors their own press centre I would limit the information I receive and would not get back all the things that potentially interest me.
Another source of incoherence is a mismatch between the filter and the keywords you have used to generate information flows from search engines (query sources). In the example below, I have created an information flow using the keyword “barclays”. If I also want to get information about Lloyds TSB, The Royal Bank of Scotland, and HSBC I will need to add these sources to the monitoring.
4. Check for focus: are my keywords well-placed in the filter
Some of the information you decide not to keep will contain your search terms, but perhaps only at the end of the article, or two important words are a long way apart from each other. There are two ways to deal with this.
To make sure that your keywords are at the start (first 25%, including the title) of a piece of information use begin: and to specify that you would like the keyword in the title of the information use title: For example:
intitle:Barclays AND begin:"online payment"
will bring back results that mention Barclays in the title and also mention online payment in the first 25%.
To indicate that you would like to have certain search terms close to each other use the ~. This will help you to make sure that the words are in the right context in the article. For example:
“Barclays online payment”~15
will bring back results that mention these three words together in a total of 15 words, roughly the same length as a sentence. You can use 25 or 40 to indicate you would like the words in the same paragraph.
Both of these two techniques will help you to reduce the quantity of information that contains your search terms but is not relevant to your needs.