The logical extension of "on-line" meta-motors is software that is installed on your computer, and that in theory nullifies the disadvantages mentioned above. The principle is exactly the same. For each request, they query several search tools and eliminate duplicate results. But what makes them original is that you do not need to be connected to the Internet to make them work. You first define your search strategy "off-line " by choosing the relevant keywords and the tools that you wish to use for each search mission. You can also define the date, time, and the frequency of your search, as well as where you want the agent to store the search results. Then, the agent spontaneously activates a connection to the Internet, queries the search tools with the keywords that you gave it, combines the results by eliminating the duplicates, saves them on your hard disk, and finally disconnects from the network to rest for a moment...
You can then view the results without having to be connected to the network, which means you have almost immediate display as all the data has been stored on your computer. The advantages of such tools quickly become clear. First, unlike on-line some meta-motors, you can freely select the engines to be used from increasingly exhaustive lists. You will not be obliged to use AltaVista with Yahoo any more! Some agents let you add any search tool, even if it is not included in their list of choices. Also, many of them allow you to use complex operators to undertake advanced searches (the agent is automatically given the responsibility of adapting the Boolean operators for each search engine). However, they do not enable you to take into account the specificities of each engine.
Naturally, they also have their limits. Like on-line meta-motors, the quantity of "noise" that they download is sometimes overwhelming. The disadvantage here is that all this irrelevant data is likely to clutter up your computer very quickly, creating the need for additional spring-cleaning on a regular basis. The "noise" will be even more significant if you select search tools without knowing their characteristics. For instance, somebody who does not know the specificities of AltaVista and Yahoo, and who decides to use them simultaneously with an agent, risks being submerged by a lot of useless information. It is therefore very important to know each search engine and directory very well before using it with an agent. They are complimentary, not competing, tools. A search agent is only a "hollow shell" without the various search tools that it is intended to query. Also, to avoid the risk of finding useless pages, it is advisable to choose your search keywords very carefully. Ideally, it is a good idea to test the effectiveness of your keywords on the Internet with an engine like HotBot or AltaVista, before launching an agent. In every case, the use of agents must be prepared and planned. Searching aimlessly will always be disappointing. All the methodological precautions that are important for effectively carrying out a search on the Internet keep their value in this context.
Another disadvantage is their slow working speed. For a complex search, an agent can take several hours to complete its work since it has to query several engines, eliminate the duplicates, connect to the various Web sites, save all the documents, and possibly process the results using your computer.
Currently, one can find many off-line meta-motors on the market (Copernic, Bullseye...).