[alicebot-archcomm] Matching Order Revised

[Sandro Golinelli]

hello there,

im just thinking about the current matching order....

as you can see in the current implementation:

<input>    _ <that> THAT <topic> TOPIC
<input>    _ <that> THAT <topic>   *
<input>    _ <that>  * <topic> TOPIC
<input>    _ <that>  * <topic>   *
<input> PATTERN <that> THAT <topic> TOPIC
<input> PATTERN <that> THAT <topic>   *
<input> PATTERN <that>  * <topic> TOPIC
<input> PATTERN <that>  * <topic>   *
<input>    * <that> THAT <topic> TOPIC
<input>    * <that> THAT <topic>   *
<input>    * <that>  * <topic> TOPIC
<input>    * <that>  * <topic>   *

..the user input is the most important piece of the conversation whatever we
are using that or topic categories.

One thing that i can't recognize in the current matching order as "usefull"
is the use of topic.

It is somewhat strange to set a topic but have a match outside of it,
just because another user input (a pattern nor a category in the Dirk's
way ) outside of the current topic could be (in the current matching order)
"more important" than some other inside the topic.

With slight modification, the matching order here below can achieve more
contextual capabilities by still pose the user input as the most important
piece of the conversation but, IMO, it shall keep the context because it's
ordered by the topic and the that, as you can see:

<input>    _ <that> THAT <topic> TOPIC
<input> PATTERN <that> THAT <topic> TOPIC
<input>    * <that> THAT <topic> TOPIC
<input>    _ <that>  * <topic> TOPIC
<input> PATTERN <that>  * <topic> TOPIC
<input>    * <that>  * <topic> TOPIC
<input>    _ <that> THAT <topic>   *
<input> PATTERN <that> THAT <topic>   *
<input>    * <that> THAT <topic>   *
<input>    _ <that>  * <topic>   *
<input> PATTERN <that>  * <topic>   *
<input>    * <that>  * <topic>   *

so as a final matching order with contextual capabilities i can think of
this:

<topic> TOPIC <that> THAT <input>    _
<topic> TOPIC <that> THAT <input> PATTERN
<topic> TOPIC <that> THAT <input>    *
<topic> TOPIC <that>  * <input>    _
<topic> TOPIC <that>  * <input> PATTERN
<topic> TOPIC <that>  * <input>    *
<topic>   * <that> THAT <input>    _
<topic>   * <that> THAT <input> PATTERN
<topic>   * <that> THAT <input>    *
<topic>   * <that>  * <input>    _
<topic>   * <that>  * <input> PATTERN
<topic>   * <that>  * <input>    *

With this matching order, i see there are some advantages.
The context is kept because, before the user input match, the topic and the
that are taken into consideration first.
Moreover, half of the matches can be skipped if no topic is set, and a
quarter of the matches can be skipped too if no that is set.

This view seems IMHO more logic and contextual than the current one.
(it's almost already there as you don't have to add a new things & tags ;-))


Looking forward to your thoughts.

Sandro

P.S.
if the matchings order are screwed up in the e-mail,
it's also available in RTF here:
http://geocities.com/pons18/Download/matching-order.rtf