基于文档值分布的Solr功能

Question

我今天正在尝试重新排序查询。

我的features.json如下：

[  
{
"name" : "documentRecency",
"class" : "org.apache.solr.ltr.feature.SolrFeature",
"params" : {
"q" : "{!func}recip( ms(NOW,timestamp), 3.16e-11, 1, 1)"
}
},
{
"name" : "textLengthScore",
"class" : "org.apache.solr.ltr.feature.SolrFeature",
"params" : {
"q" : "{!func}recip(rord(scale(textLength, 0, 1)), 1,1000,1000)"
}
},
{
"name" : "numCategoriesScore",
"class" : "org.apache.solr.ltr.feature.SolrFeature",
"params" : {
"q" : "{!func}recip(rord(scale(numCategories, 0, 1)), 1,1000,1000)"
}
},
{
"name" : "numSectionsScore",
"class" : "org.apache.solr.ltr.feature.SolrFeature",
"params" : {
"q" : "{!func}recip( rord(scale(numSections, 0, 1)), 1,1000,1000)"
}
},
{
"name" : "numLinksScore",
"class" : "org.apache.solr.ltr.feature.SolrFeature",
"params" : {
"q" : "{!func}recip( rord(scale(numLinks, 0, 1)), 1,1000,1000)"
}
},
{
"name" : "originalScore",
"class" : "org.apache.solr.ltr.feature.OriginalScoreFeature",
"params" : {}
}
]

我的model.json看起来是这样的：

{
"class" : "org.apache.solr.ltr.model.LinearModel",
"name" : "myModel",
"features" : [
{ "name" : "documentRecency" },
{ "name" : "textLengthScore" },
{ "name" : "numCategoriesScore" },
{ "name" : "numSectionsScore" },
{ "name" : "numLinksScore" },
{ "name" : "originalScore" }
],
"params" : {
"weights" : {
"documentRecency" : 0.2,
"textLengthScore" : 0.5,
"numCategoriesScore" : 0.3,
"numSectionsScore": 0.6,
"numLinksScore" : 0.4,
"originalScore" : 0.3
}
}
}

当重新排列我的结果时，我得到以下错误：

"java.lang.RuntimeException: Exception from createWeight for SolrFeature [name=textLengthScore, params={q={!func}recip(rord(scale(textLength, 0, 1)), 1,1000,1000)}] Failed to parse feature query."

这个查询有什么问题？我正试图得到一个"；textLengthScore"；通过从0,1缩放所有文本长度，然后像在documentRecency中一样对其进行评分。

哦，是的，这是schema.xml:

<?xml version="1.0" encoding="UTF-8"?>
<schema name="sem" version="1.6">
<uniqueKey>id</uniqueKey>
<fieldType name="string" class="solr.StrField"/>
<fieldType name="longstring" class="solr.TextField">
<analyzer type="query">
<tokenizer class="solr.StandardTokenizerFactory"/>
<filter class="solr.LowerCaseFilterFactory"/>
<filter class="solr.SnowballPorterFilterFactory" language="English"/>
<filter class="solr.SynonymFilterFactory" synonyms="synonyms.txt" ignoreCase="true" expand="true" tokenizerFactory="solr.StandardTokenizerFactory"/>
</analyzer>
<similarity class="solr.ClassicSimilarityFactory"/>
</fieldType>
<fieldType name="tstamp" class="solr.DatePointField"/>
<fieldType name="number" class="solr.TrieIntField" precisionStep="0" docValues="true"/>
<field name="id" type="string" indexed="true" required="true" stored="true"/>
<field name="title" type="longstring" indexed="true" required="true" stored="true"/>
<field name="timestamp" type="tstamp" indexed="true" required="true" stored="true"/>
<field name="categories" type="longstring" indexed="true" multiValued="true" stored="true"/>
<field name="text" type="longstring" indexed="true" multiValued="true" stored="true"/>
<field name="fullText" type="longstring" indexed="true" stored="true"/>
<field name="links" type="longstring" indexed="true" multiValued="true" stored="true"/>
<field name="textLength" type="number" indexed="true" stored="true"/>
<field name="numCategories" type="number" indexed="true" stored="true"/>
<field name="numLinks" type="number" indexed="true" stored="true"/>
<field name="numSections" type="number" indexed="true" stored="true"/>
</schema>

Answer 1

我认为错误来自rord()，只需要一个字段名作为参数。

在使用rord((或ord((之前，您不需要缩放值，缩放不会影响它们的字典顺序，因此每个值索引都将保持不变。

换句话说，"{!func}recip(rord(textLength), 1,1000,1000)"应该是好的。