Tweaking of search engine results and news titles based on the user's interests and preferences (patent-pending).

This invention presents a client computer based system and a way to automatically resort the search results received from a search engine, and the news titles received from several news sources, by moving the results that are more relevant to this particular computer user to a better position on the list.

The problem: The primary problem addressed by this invention relates to how searches are currently performed on the Internet. While processing a search request, a search engine relies only on the information that it has on its (server) side and knows nothing about the particular user that is performing the search.

Let's assume that we have two music fans; Anna and Angelina.
Anna loves pop music. She visits pop music web sites, reads pop music news and discusses the latest pop music news in her personal blog. Angelina loves classical music. She visits classical music web sites, exchanges links of free classical music web sites with her friends and reads articles about Sebastian Bach and Beethoven.
When next time they perform a search with say 'buy DVD album' or 'music events' or 'download MP3' keywords they both will receive the same set of results although they expect to see different kind of links.

The solution: In order to study the user’s interests and preferences we need to install a special software program on to the user's computer that will analyze the words and phrases that the user prefers to read on web sites, eBooks, write in chat programs, web forms and etc. Each word will receive a rank depending on how often it is used and whether the user types it or just reads it. After some time studying the user, we will have a complete database of words and phrases that will be specific for this particular computer user (words connected to his/her job or hobby). The collected information must be stored into properly encrypted database in order to prevent the leak of user's privacy.

When next time the user performs a search, our software will send a request to the search engine and then will receive a list, say of 100 results. Then our software will download the text for each of these 100 web pages (in order to avoid the overusage of the user's network traffic the search engine may reply with a zipped XML file that will already contain the cached text of all of these 100 web pages). Then our software program will analyze the text of each web page to see how relevance it is to the user's interests and preferences. After this it will resort the original list by moving the results that are more relevant to this particular computer user to a better position on the list and then it will show the processed list to the user.

The same tasks can be performed for news titles. After receiving the list of news titles (for example, from RSS/Atom news feeds), our software component will download the full text for each news story, analyze this text, and move the news that are more relevant to this particular computer user to a better position on the list.

Example #1: The user is a UAB basketball team fan. He/she reads a lot about this team and about the team players, discuses them in specialized news groups and forums, writes about them in his/her personal blog, etc. In the result of the word 'basketball', the name of this team and names of team players will be included in our database and will have a good rank. Next time he/she will search for some information about the sport, sport events, sport equipment etc, the web pages containing any information about this team will be placed at a higher position by our software program even if the search engine initially placed these web pages at a lower position.

Example #2: The user loves pop music, reads articles about his/her favorite group, writes some comments about them in forums and blogs, buys their DVD albums etc. In this case the group name, artist names, song names etc. will be included in our database and these words will have a good rank. If this user has subscribed to several RSS news feeds, then all news that contain any information about the group will be automatically placed over other news stories.