Advanced search agents
Certain search agents offer some very interesting possibilities. The function common to all the agents previously described is that they query only the engines that were selected, based on keywords chosen by the user. However, a certain number of companies have developed additional functions.
Personalization by industry branch.
- Sectoral engines are tools specialized in just one precise field (finance, insurance, education, literature, etc). They are often very rich independent databases that cannot be reached by general search robots (what the specialists call The Invisible Web). They remain very useful when the traditional tools fail. However, the fact that they are scattered makes their use often rather long and difficult. For this reason, certain search agents are specialized by industry branch. One of the most effective, WebSeeker, integrates several hundred search engines which it classifies in many categories (art, cars, business, computers, education, employment, finance, leisure, theatre, health, homes, multi-media, etc). It also makes it possible to create personalized categories. Citeline offers a very advanced agent in the field of health. We should also point out two very effective search agents : Copernic and Strategic Finder from DigiMind.
- Monitoring, filtering and archiving of results. Some agents (BullsEye, WebSeeker, etc.) allow you to track search results over time. In other words, at regular intervals, they repeat the query with all the search tools that you selected. You can therefore follow the evolution of the results. Moreover, it is also possible to filter these results to highlight the differences over time, possibly using new keywords. It is often possible to archive the results of each search.
- Linguistic analysis of search queries. Normally, to launch a search agent you must indicate the keywords that you wish to search for, as well as the tools that you want it to use (HotBot and AltaVista, but not InfoSeek, etc). You must therefore deal with a certain number of sometimes tiresome constraints. For this reason, certain agents accept queries in natural language, i.e. in full sentences and not in the form of keywords (DigOut4U). hen, through a linguistic module, the agent translates your personal formulations by extracting the most significant keywords. In addition, you often do not even have to choose which directory or search engine to use. The agent queries a certain number by default. At the time of writing, the results are still a little mixed, as the linguistic adaptation is not very powerful yet.
- Analysis of the results. Certain agents perform advanced processing of the information collected. They can be very diverse in nature and can imply higher costs. SemioMap for example, based on the principles of semiotics, makes it possible to obtain a precise cartography of downloaded information in order to correctly understand the present situation and to detect possible "weak signals". Umap from Trivium adopts a similar approach but with a different technology. Other agents carry out semantic indexing of the results, or can automatically generate summaries for each site. DigOut4U is the only tool that downloads the semantic contents of Web pages, as well as the contents of newsgroups in a continuous mode (all of the hypertext links that are found are analyzed ad infinitum!). The results vary greatly according to the technologies used. It is clear that an agent costing $20 will not be able to obtain the same results as an agent architecture that costs several thousands of dollars.
As we have already said, it is a good idea to master the different search engines before deciding to use these agents. In addition, if you are looking for a person, a company, a product, software or any other information, and that you do not wish to use each search tool manually, you can still use the standard search methodology. It often adapts very well to search agents. As an example, if you want to obtain a maximum amount of information on a new product launched by one of your competitors, you can select most powerful search engines within your agent (HotBot, AltaVista, InfoSeek) as well as a certain number of sectoral engines.