NEO4J Cypher查询速度慢

Question

我努力优化以下查询，但无法将其降低到两分钟以下 - 这是预期的吗？有什么方法可以加快速度。我添加了索引并尽可能使用WITH。

match (md:MarketingDetail {Status: 'Live'})
with md limit 10
match (md) -[:Has_Area]-> (a:Area)
with md, a
match (ss:SavedSearch) -[:Has_Area]->(a)
with md, ss match
(md) -[:Has_Trade] -> (st:SectorTrade) <-[:Has_Trade]- (ss)
where ((md.FreeholdTenure ='Freehold'
and ss.FreeholdTenure = 'true'
and (md.FreeholdSearchPrice >= ss.PriceFrom
or md.FreeholdSearchPrice is null)
and (md.FreeholdSearchPrice <= ss.PriceTo
or md.FreeholdSearchPrice is null))
or (md.LeaseholdTenure is not null
and ss.LeaseholdTenure = 'true'
and (md.LeaseholdSearchPrice >= ss.PriceFrom
or md.LeaseholdSearchPrice is null)
and (md.LeaseholdSearchPrice <= ss.PriceTo
or md.LeaseholdSearchPrice is null)))
return count(ss)

以下是上述查询的个人资料 -

谢谢！

Answer 1

在不了解您的图表及其结构/大小的情况下，很难确切知道如何有效地查询它。在查询所需模式之前，您的查询似乎与每个md和ss组合相匹配，根据图表，这可能是也可能不是最好的方法，但我已尝试过替代下面。

您可以在查询之前放置PROFILE以查看其执行方式并查找瓶颈。您可以通过运行:schema ??

来发布此结果以及索引

MATCH (md:MarketingDetail {Status: 'Live'})
WHERE 
  (
    md.FreeholdTenure ='Freehold'
    OR md.LeaseholdTenure is not null
    OR md.LeaseholdSearchPrice is null
    OR md.LeaseholdSearchPrice is null
  )
WITH md LIMIT 10
match (md) -[:Has_Area]-> (a:Area)<-[:Has_Area]-(ss:SavedSearch)-[:Has_Trade]->(st:SectorTrade)-[:Has_Trade]->(md)
where
  AND
  (
    md.FreeholdTenure ='Freehold'
    and ss.FreeholdTenure = 'true'
    and (md.FreeholdSearchPrice >= ss.PriceFrom or md.FreeholdSearchPrice is null)
    and (md.FreeholdSearchPrice <= ss.PriceTo or md.FreeholdSearchPrice is null)
  )
  or (
    md.LeaseholdTenure is not null
    and ss.LeaseholdTenure = 'true'
    and (
      md.LeaseholdSearchPrice >= ss.PriceFrom
      or md.LeaseholdSearchPrice is null
    )
    and (
      md.LeaseholdSearchPrice <= ss.PriceTo
      or md.LeaseholdSearchPrice is null
    )
  )  
return count(ss)

Answer 2

根据您的PROFILE，您没有:MarketingDetail(Status)的索引，这对您的查询非常重要，因为第一个MATCH需要这个索引。

此外，重构您的查询（可能以@DonWeldon建议的方式）应该会有所帮助。