kvindex的查询语法

[chendaxixi]

kvindex的查询语法确认： 原先的kvindex(f_index_pr实现的)的查询语法为: select kvindex(queryPath, patternPath, pattern starttime, pattern endtime, epsilon (,alpha, beta)) from queryPath WhereClause 之后和向东学长讨论的查询语法有：

1. select ... from query path where kvmatch(pattern path, ...) and ...
2. select kvindex(pattern path,...) from queryPath WhereClause 需要确定下来新的语法并更新f_index_pr分支么

[wujysh]

@kr11 @jixuan1989

If we decide to integrate KV-match of ICDE 2019 to IoTDB (according to the Apache incubation proposal), it's better to support multiple (at least 4) types of queries with one single index (RSM-ED, RSM-DTW, cNSM-ED, cNSM-DTW).

So the syntax of the query statement should be robust enough to support all of them. Following is the initial plan in my opinion.

SELECT kv-match(<pattern_path>, <parameter_1[,parameter_2,...]>) FROM <query_path> WHERE <time_criterion>

<pattern_path>

It's better it can receive a sub-query with where clause on time, to use a subsequence as pattern.

<parameter_1[,parameter_2,...]>

RSM-ED: SELECT kv-match(<pattern_path>, "RSM-ED", <epsilon>) FROM <query_path> WHERE <time_criterion>

RSM-DTW: SELECT kv-match(<pattern_path>, "RSM-DTW", <epsilon>, <rho>) FROM <query_path> WHERE <time_criterion> Where <rho> is the constraint on the warping path of DTW measure.

cNSM-ED: SELECT kv-match(<pattern_path>, "cNSM-ED", <epsilon>, <alpha>, <beta>) FROM <query_path> WHERE <time_criterion> Where <alpha> and <beta> are the constraints on mean value and standard deviation differences (refer to the paper).

cNSM-DTW: SELECT kv-match(<pattern_path>, "cNSM-DTW", <epsilon>, <rho>, <alpha>, <beta>) FROM <query_path> WHERE <time_criterion> Where <rho> is the constraint on the warping path of DTW measure, and <alpha> and <beta> are the constraints on mean value and standard deviation differences (refer to the paper).

If you have any plan or suggestion on the integration, please let me know.