|
|
 | | From: | Yao Ziyuan | | Subject: | Input/Output Cache Systems for the Web | | Date: | Sat, 22 Jan 2005 11:30:12 +0800 |
|
|
 | A typical Internet user's information search activity can be
characterized as:
1. Give some Input to a search engine (e.g. search keywords);
2. Validate the search results;
3. Find out the right Output among the search results.
Step 2 above can be time-consuming. If we have a network of cache
databases that store such [Input] / [Output] pairs as database records,
then only the first information searcher needs to do the hard work of
step 2, and subsequent searchers for the same/similar Input can directly
get the Output from such cache database networks. Maintenance of such
database networks can borrow ideas from the establishment of the USENET.
Example applications:
1) A software download link information system. Each database record can be:
[Software Name] [Additional Keywords] [Download Links] [Last Updated].
The first searcher who successfully finds a download link for a certain
software product can submit this (Software Name, Download Link) data
pair to such an information system, and subsequent searchers for the
same software can directly get the result (Download Link) from such
download link information system.
2) A source language to target language phrasebook system. The mechanism
is the same as above except that each record is:
[Source Language Phrase] [Target Language Phrase] [Comment].
The application of Example 1 was proposed by Ke Xiaofeng, and that of
Example 2 was proposed by Yao Ziyuan.
Yao Ziyuan
|
|
| | From: | Yao Ziyuan | | Subject: | Re: Input/Output Cache Systems for the Web | | Date: | Sat, 22 Jan 2005 11:53:55 +0800 |
|
|
| Records that can't be well described by a single input keyphrase (such
as a patent claim) can be stored and retrived with a hierarchy of
classifications.
Wiki can be an underlying platform for such collaborative information
systems, but the data submission process could be more convenient or
even automated. For example, a file downloading program should
automatically submit the (Software Name, Download Link) information to
an online download link repository if it successfully begins downloading.
Yao
Yao Ziyuan wrote:
> A typical Internet user's information search activity can be
> characterized as:
>
> 1. Give some Input to a search engine (e.g. search keywords);
> 2. Validate the search results;
> 3. Find out the right Output among the search results.
>
> Step 2 above can be time-consuming. If we have a network of cache
> databases that store such [Input] / [Output] pairs as database records,
> then only the first information searcher needs to do the hard work of
> step 2, and subsequent searchers for the same/similar Input can directly
> get the Output from such cache database networks. Maintenance of such
> database networks can borrow ideas from the establishment of the USENET.
>
> Example applications:
>
> 1) A software download link information system. Each database record can
> be:
>
> [Software Name] [Additional Keywords] [Download Links] [Last Updated].
>
> The first searcher who successfully finds a download link for a certain
> software product can submit this (Software Name, Download Link) data
> pair to such an information system, and subsequent searchers for the
> same software can directly get the result (Download Link) from such
> download link information system.
>
> 2) A source language to target language phrasebook system. The mechanism
> is the same as above except that each record is:
>
> [Source Language Phrase] [Target Language Phrase] [Comment].
>
> The application of Example 1 was proposed by Ke Xiaofeng, and that of
> Example 2 was proposed by Yao Ziyuan.
>
> Yao Ziyuan
|
|
| | From: | yaoziyuan at gmail.com | | Subject: | Re: Input/Output Cache Systems for the Web | | Date: | 22 Jan 2005 13:34:05 -0800 |
|
|
| For example application #1 (file url sharing), it's already reality. A
download manager FlashGet has started a website http://www.shareurl.com
which tells you what download links other FlashGet users use to
download the file of a given filename. Although it would be better to
search by product name, the current design is automatic and doesn't
require a user to manually submit shared information.
Yao Ziyuan wrote:
> A typical Internet user's information search activity can be
> characterized as:
>
> 1. Give some Input to a search engine (e.g. search keywords);
> 2. Validate the search results;
> 3. Find out the right Output among the search results.
>
> Step 2 above can be time-consuming. If we have a network of cache
> databases that store such [Input] / [Output] pairs as database
records,
> then only the first information searcher needs to do the hard work of
> step 2, and subsequent searchers for the same/similar Input can
directly
> get the Output from such cache database networks. Maintenance of such
> database networks can borrow ideas from the establishment of the
USENET.
>
> Example applications:
>
> 1) A software download link information system. Each database record
can be:
>
> [Software Name] [Additional Keywords] [Download Links] [Last
Updated].
>
> The first searcher who successfully finds a download link for a
certain
> software product can submit this (Software Name, Download Link) data
> pair to such an information system, and subsequent searchers for the
> same software can directly get the result (Download Link) from such
> download link information system.
>
> 2) A source language to target language phrasebook system. The
mechanism
> is the same as above except that each record is:
>
> [Source Language Phrase] [Target Language Phrase] [Comment].
>
> The application of Example 1 was proposed by Ke Xiaofeng, and that of
> Example 2 was proposed by Yao Ziyuan.
>
> Yao Ziyuan
|
|
|
