| Last Modified On : | September 26, 2007 1:13 AM PDT |
Rate |
|
In this section we describe how Intel Mash Maker actually works. Normal users will probably want to skip this chapter.
The diagram below summarises the communication structure for Intel Mash Maker:
The Mash Maker client communicates with machines on the internet in the following ways:
- Mash Maker Database: The client talks to the mash maker database in order to metadata with other users. This metadata consists of extractors, mashups, and information about which mashups work well for different kinds of page.
- Web Sites: The client talks to web sites to retreive content that can be combined with the current page
- Widget Services: Mash Maker widgets may communicate with a home server in order to provide their user interface.
The Intel Mash Maker database contains the following tables:
- Users: login details for each user
- Extractors: how to understand different kinds of page.
- Mashups: mashups that can be applied to content
- Scoring: which users applied which mashups to which pages
Note that the database does not store any of the actual data that the client manipulates. It only contains metadata.
The Mash Maker client sends the following messages to the database:
- Retrieve Mashup Database: Currently the entire database is downloaded.
- Retrieve Extractors for a Domain: Download all extractors applicable to pages in a given domain (e.g. bbc.co.uk). No login details and sent, to avoid relealing potentially sensitive information about which domains a user visits.
- Score Mashup: Let the database know that the user either used a mashup, or marked it as good/bad
- Edit Extractor: Edit/create an extractor
- Edit Mashup: Edit/create a mashup
Mashup suggestions are computed on the client rather than on the server, since the suggestions may be based on private information, such as the content of the current page, and the content of other pages the user has recently browsed.
