与搜索混淆：建议

发布于 2024-12-02 06:18:46 字数 825 浏览 0 评论 0原文

我无法借助 search:suggest 函数提供这个简单的自动完成功能。

基于 Marklogic 演示数据中的奥斯卡语料库，我尝试提供一个建议查询，例如，即使用户当前正在写入“Robert Lo”或“Loggia Rob”甚至“，也能够返回“Robert Loggia”作为答案。 L·罗布”。

目前，我只能通过这个简单的查询在他的一些同行中返回“Robert Loggia”：

let $options := 
<options xmlns="http://marklogic.com/appservices/search">
  <default-suggestion-source>
    <range type="xs:string">
      <element ns="http://marklogic.com/wikipedia" name="name" />
    </range>
  </default-suggestion-source>
</options>
return search:suggest("Rob",$options)

但是一旦我写了一个两个词的短语，我不确定如何写它，因为没有人知道答案与我正在等待的内容是正确的。例如：

search:suggest("Robert Lo",$options)

或者

search:suggest(("Robert", "Lo"),$options)

是否由于缺少选项、索引配置错误或误用功能而导致？

感谢您的帮助

原文

I'm not able to provided this simple autocompletion feature with the help of the search:suggest function.

Based on the Oscars corpus from Marklogic Demo data, I try to provide a suggest query which was for instance able to return "Robert Loggia" as an answer even if the user is currently writing "Robert Lo" or "Loggia Rob" or even "L Rob".

Currently, I'm only able to return "Robert Loggia" amongst some of his peers with this simple query :

let $options := 
<options xmlns="http://marklogic.com/appservices/search">
  <default-suggestion-source>
    <range type="xs:string">
      <element ns="http://marklogic.com/wikipedia" name="name" />
    </range>
  </default-suggestion-source>
</options>
return search:suggest("Rob",$options)

But as soon as I'm writing a two words phrase, I'm not sure to understand how to write that, because none of the answer is correct with what I'm waiting. For instance :

search:suggest("Robert Lo",$options)

search:suggest(("Robert", "Lo"),$options)

Is it due to missing option, indexes misconfiguration or a misused feature ?

Thanks For Help

分享到QQ

分享到微博

如果你对这篇内容有疑问，欢迎到本站社区发帖提问参与讨论，获取更多帮助，或者扫码二维码加入 Web 技术交流群。

发布评论

需要登录才能够评论，你可以免费注册一个本站的账号。

我一直都在从未离去 2024-12-09 06:18:46

它具有语法意识。

如果您想获得有关该短语的建议，则需要在该短语周围添加引号，否则它将单独处理标记。请尝试以下操作：

search:suggest('"Robert Lo"',$options)

It's grammar-aware.

You need to add quotes around the phrase if you want to get suggestions on the phrase, otherwise it treats the tokens separately. Try the following:

search:suggest('"Robert Lo"',$options)

回复收藏 0 原文

深海里的那抹蓝 2024-12-09 06:18:46

这是一个常见的错误，我自己也犯过好几次。

正如科琳所说，您需要在查询字符串两边加上引号。这与一般使用搜索 API 或 google 进行搜索相同：带引号的参数被视为一个短语，没有每个标记的都是一个单独的参数。

如果没有引号，默认行为是将这两个术语视为单独的标记，并在查询中对它们进行 AND 或 OR 运算，具体取决于您的配置。

回复收藏 0 原文

青春有你 2024-12-09 06:18:46

在 OS X 上使用 4.2-6 对我来说也很奇怪。这是我的测试结果，使用完整的奥斯卡数据集。

import module namespace search = "http://marklogic.com/appservices/search"
    at "/MarkLogic/appservices/search/search.xqy";

let $options := 
<options xmlns="http://marklogic.com/appservices/search">
  <default-suggestion-source>
    <range type="xs:string">
      <element ns="http://marklogic.com/wikipedia" name="name" />
    </range>
  </default-suggestion-source>
</options>
return search:suggest("Robert Lo", $options)
=>
Robert "Loretta Young"
Robert "Lorraine Bracco"
Robert "Lotte Lenya"
Robert "Louis Calhern"
Robert "Louis Gossett, Jr."
Robert "Louis Malle"
Robert "Louise Fletcher"

看起来这些值是为 or-query(('Robert', 'Lo')) 生成的，然后以奇怪的方式合并。如果我们直接使用 cts lexicon 函数，事情看起来很正常。

cts:element-value-match(
  QName('http://marklogic.com/wikipedia', 'name'),
  'Robert Lo*')
=>
Robert Loggia

单词交换方面使事情变得更加复杂，并且超出了自动建议通常所做的范围（以及 search:suggest 和 cts:element-value-match 可以做的事情没有帮助）。

这是一个使用额外的字符串操作来创建两个模式，处理单词对的版本。它假设第一个单词始终是完整的，而第二个单词可能是存根，但可能需要交换这些单词以进行匹配。如果您定义了所需的逻辑，您应该能够对其进行修改以满足您的要求。请注意，函数映射正在发挥作用，因此在评估期间每个模式都会调用一次 cts:element-value-match 。这可能会对性能产生影响。

let $toks := cts:tokenize($INPUT)[. instance of cts:word]
let $pat := (
  if (count($toks) eq 1) then concat($toks, '*')
  else (
    concat($toks[2], '* ', $toks[1]),
    concat($toks[1], ' ', $toks[2], '*')))
return cts:element-value-match(
  QName('http://marklogic.com/wikipedia', 'name'),
  $pat)
=>
Robert Loggia

It looks odd to me too, using 4.2-6 on OS X. Here's what the results looked like in my test, using the full oscars data set.

import module namespace search = "http://marklogic.com/appservices/search"
    at "/MarkLogic/appservices/search/search.xqy";

let $options := 
<options xmlns="http://marklogic.com/appservices/search">
  <default-suggestion-source>
    <range type="xs:string">
      <element ns="http://marklogic.com/wikipedia" name="name" />
    </range>
  </default-suggestion-source>
</options>
return search:suggest("Robert Lo", $options)
=>
Robert "Loretta Young"
Robert "Lorraine Bracco"
Robert "Lotte Lenya"
Robert "Louis Calhern"
Robert "Louis Gossett, Jr."
Robert "Louis Malle"
Robert "Louise Fletcher"

It looks like the values are generated for or-query(('Robert', 'Lo')) and then merged in an odd way. If we drop down to using the cts lexicon function, things look normal.

cts:element-value-match(
  QName('http://marklogic.com/wikipedia', 'name'),
  'Robert Lo*')
=>
Robert Loggia

The word-swap aspect makes things more complicated, and goes beyond the scope of what autosuggest usually does (and what search:suggest and cts:element-value-match can do without help).

Here's a version that uses extra string manipulation to create two patterns, handling pairs of words. It assumes that the first word is always complete, while the second might be a stub, but that the words may need to be swapped for a match. If you define your desired logic, you should be able to modify this to suit your requirements. Note that function mapping is in play, so cts:element-value-match is called once per pattern during evaluation. That may have a performance impact.

let $toks := cts:tokenize($INPUT)[. instance of cts:word]
let $pat := (
  if (count($toks) eq 1) then concat($toks, '*')
  else (
    concat($toks[2], '* ', $toks[1]),
    concat($toks[1], ' ', $toks[2], '*')))
return cts:element-value-match(
  QName('http://marklogic.com/wikipedia', 'name'),
  $pat)
=>
Robert Loggia

回复收藏 0 原文

~没有更多了~