Question

我有一个包含3列的简单文档集：

EventID（int，key）
EventName（字符串，可搜索）
CategoryName（字符串，可搜索）

我提出了以下要求：

https://<service_name>.search.windows.net/indexes/<index_name>/docs?api-version=2016-09-01&search=basketball

并收到以下输出：

{
"value": [
    {
        "@search.score": 2.289929,
        "EventID": "1186",
        "EventName": "Basketball RACK Event creation from PT zone",
        "CategoryName": "Basketball"
    },
    {
        "@search.score": 1.8826654,
        "EventID": "1000",
        "EventName": "BTNtest",
        "CategoryName": "Basketball"
    },
    {
        "@search.score": 1.8826654,
        "EventID": "562",
        "EventName": "BTN TEST (former TT Test)",
        "CategoryName": "Basketball"
    },
    {
        "@search.score": 1.7038364,
        "EventID": "834",
        "EventName": "POSTROLL TEST 2",
        "CategoryName": "Basketball"
    },
    {
        "@search.score": 1.7038364,
        "EventID": "964",
        "EventName": "MLB test",
        "CategoryName": "Basketball"
    },
    ...
    {
        "@search.score": 1.238302,
        "EventID": "31",
        "EventName": "Monday Event New",
        "CategoryName": "Basketball"
    },
    {
        "@search.score": 1.0656936,
        "EventID": "1151",
        "EventName": "sof-632 EST past event",
        "CategoryName": "Basketball"
    }
]

}

编辑：输出对我来说并不明显。除第一项之外的所有文档与类别字段中的搜索请求“Basketball”完全匹配，但其分数从1.88到1.06不等。因此，当我添加具有新鲜度功能的评分配置文件以按日期对项目进行排序时，输出变得非常奇怪 - 我看到来自同一类别篮球的文档，但其顺序是意外的。在这种情况下如何才能提高搜索质量？

如何抑制Azure搜索评分的不良行为

0 个答案: