User talk:Eggfree: Difference between revisions
| Line 3: | Line 3: | ||
This page catalogs a set of design ideas and proposals for some current projects. I include a synopsis below each. | This page catalogs a set of design ideas and proposals for some current projects. I include a synopsis below each. | ||
'''Bookmarking''' Various research leans to the idea that effective bookmark lists can be imputed and modified based on a users browsing habits then proposed, rather than requiring explicit actions. | '''Bookmarking''' Various research leans to the idea that effective bookmark lists can be imputed and modified based on a users browsing habits then proposed, rather than requiring explicit actions. Thee are three steps in this bookmarking scheme: bookmark proposal, presentation, and editing. | ||
=== Proposing bookmarks=== | === Proposing bookmarks=== | ||
Proposal is the first st | |||
====Analysis==== | |||
*1) analyze browsing history to cluster page groups that are often browsed in sequence (for example I often view 'slashdot','nytimes',theinquirer' first thing in the morning) | *1) analyze browsing history to cluster page groups that are often browsed in sequence (for example I often view 'slashdot','nytimes',theinquirer' first thing in the morning) | ||
*2) score pages in the current history by frequency of visitation and length of stay. | *2) score pages in the current history by frequency of visitation and length of stay. | ||
| Line 11: | Line 14: | ||
**http://www.nytimes.com, | **http://www.nytimes.com, | ||
**http://www.nytimes.com/2005/11/15/international/asia/15cnd-flu.html | **http://www.nytimes.com/2005/11/15/international/asia/15cnd-flu.html | ||
all | all aggregate to score nytimes.com. | ||
====Candidate Presentation==== | |||
These first steps produce a set of likely bookmark candidates, and a set of groups. We can assume there is a list of existing bookmarks. We compare the existing bookmarks against our new proposal candidates. At the beginning of each browsing session we present these cadidates to the user for a "thumbs up" or "thumbs-down" if a user accepts the bookmark, we incorporate it into the permanent bookmark presentation. If not | |||
Revision as of 19:04, 15 November 2005
Design Topics
This page catalogs a set of design ideas and proposals for some current projects. I include a synopsis below each.
Bookmarking Various research leans to the idea that effective bookmark lists can be imputed and modified based on a users browsing habits then proposed, rather than requiring explicit actions. Thee are three steps in this bookmarking scheme: bookmark proposal, presentation, and editing.
Proposing bookmarks
Proposal is the first st
Analysis
- 1) analyze browsing history to cluster page groups that are often browsed in sequence (for example I often view 'slashdot','nytimes',theinquirer' first thing in the morning)
- 2) score pages in the current history by frequency of visitation and length of stay.
- 3) aggregate page scores for pages in the same domain that cluster:
all aggregate to score nytimes.com.
Candidate Presentation
These first steps produce a set of likely bookmark candidates, and a set of groups. We can assume there is a list of existing bookmarks. We compare the existing bookmarks against our new proposal candidates. At the beginning of each browsing session we present these cadidates to the user for a "thumbs up" or "thumbs-down" if a user accepts the bookmark, we incorporate it into the permanent bookmark presentation. If not