falcon-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Ying Zheng (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (FALCON-1095) Support free-text search for entities
Date Thu, 26 Mar 2015 21:54:53 GMT

    [ https://issues.apache.org/jira/browse/FALCON-1095?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14382765#comment-14382765
] 

Ying Zheng commented on FALCON-1095:
------------------------------------

Your understanding is correct. The current implementation by Falcon-914 is doing subsequence
matching, i.e. adding "*' after each character and before the first character. I didn't put
too much explanation because I think people already understand the well-established concept
of "subsequence matching".

Oh...I always type prefix of keywords in IntelliJ and thought it was doing exactly subsequence
matching....Forget about this wrong example. Thanks for correcting it, Scott.

Yes, as you mentioned, the user needs to enter distinguishable substrings of keywords in name
to better use this feature. We expect most of the queries contain full keywords (e.g. "house",
"texas" for entity name "houseSellTexas" in the example) to filter wired or unexpected results.

Searching by prefix of keywords may filter more unexpected results. To do this, we need to
agree with the user on a way to split keywords in the name. For example, users need to follow
camel case, or they use delimiters (e.g. "-", "_") to separate keywords. [~ajayyadava]: What's
your opinion?

> Support free-text search for entities
> -------------------------------------
>
>                 Key: FALCON-1095
>                 URL: https://issues.apache.org/jira/browse/FALCON-1095
>             Project: Falcon
>          Issue Type: New Feature
>    Affects Versions: 0.6
>            Reporter: Ying Zheng
>            Assignee: Ying Zheng
>             Fix For: 0.7
>
>
> When users have hundreds or thousands of entities to manage, they may not remember the
exact name/tag of the entity. We would like to provide a more convenient way for the user
to search an entity by providing keywords in name and tags. We also plan to provide a search
UI, similar to Visual Search, to allow the user to enter keywords in a search box and display
the filtered results. The user can click on the entity to view details and search its instances
as well.
> To support free-text search, we plan to extend the LIST API:
> GET api/entities/list/:entity-type
> Parameters changed/added:
>     1) :entity-type  Valid options are cluster, feed, process and all. Here "all" represents
feeds and processes.
>     2) nameseq     Subsequence of entity name. Not case sensitive. Example: "Sample1"
is a subsequence of entity name "SampleFeed1". Note that we rename "pattern" parameter to
"nameseq" to make it more intuitive to the user, but the functionality is already there in
Falcon-914 .
>     3) tagkey         Tag keywords, separated by comma. Not case sensitive.
> Results
>     Return the total number of entity results and a list of entities.
> e.g. falcon entity -list -nameseq housetexas -tagkey estate
> This should return the entity with name "houseSellTexas" and tags "category=realestate".



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message