In the 1992 paper Using collaborative filtering to weave an information tapestry, David Goldberg, David Nichols, Brian M. Oki and Douglas Terry describe Tapestry, an email system they developed at Xerox Parc. To help users cope with information (over)load, the system allows for collaborative filtering, i.e. users get to annotate documents (messages) to indicate their opinions (such as whether it was particularly interesting) and users also get to filter for messages based on people’s annotations and responses (such as whether they replied). The system provides a query language (Tapestry Query Language – TQL) to allow configuring of custom filters.
The paper is foundational in the literature of collaborative filtering and recommendation systems and has been quite widely cited. It has the building blocks for a working collaborative filtering system and introduces the idea of letting people make use of other people’s experiences in a very practical way.
Today’s systems are generally more evolved, requiring very little explicit direction from the end user. However, we generally also don’t know why today’s recommendation systems make the recommendations they do. The systems generally don’t go out of their way explaining themselves. I think there is definitely something to be said for being able to configure the filters yourself.