文本相似度 API 接入文档text-similarity

匿名每日 100 次、QPS 5；登录用户每日 500 次、QPS 10（全部免费）。零上游成本，纯本地计算。

获取 API Key：登录 https://apizero.cn/account/keys

以下 5 种语言示例都是可直接运行的，只需把 YOUR_API_KEY 替换为实际 Key。

curl -X POST "https://v1.apizero.cn/api/text-similarity" \
  -H "X-Api-Key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
  "text1": "今天天气不错，适合出门散步",
  "text2": "今天天气真好，适合出门走走"
}'

import requests

resp = requests.request(
    "POST",
    "https://v1.apizero.cn/api/text-similarity",
    headers={"X-Api-Key": "YOUR_API_KEY", "Content-Type": "application/json"},
    json={
    "text1": "今天天气不错，适合出门散步",
    "text2": "今天天气真好，适合出门走走",
},
    timeout=15,
)
resp.raise_for_status()
print(resp.json())

// Node.js 18+ / 浏览器原生 fetch
const res = await fetch("https://v1.apizero.cn/api/text-similarity", {
  method: "POST",
  headers: {
    "X-Api-Key": "YOUR_API_KEY",
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    "text1": "今天天气不错，适合出门散步",
    "text2": "今天天气真好，适合出门走走"
  }),
});
if (!res.ok) throw new Error(`HTTP ${res.status}`);
const data = await res.json();
console.log(data);

package main

import (
	"bytes"
	"fmt"
	"io"
	"net/http"
)

func main() {
	body := []byte(`{"text1":"今天天气不错，适合出门散步","text2":"今天天气真好，适合出门走走"}`)
	req, _ := http.NewRequest("POST", "https://v1.apizero.cn/api/text-similarity", bytes.NewBuffer(body))
	req.Header.Set("X-Api-Key", "YOUR_API_KEY")
	req.Header.Set("Content-Type", "application/json")

	resp, err := http.DefaultClient.Do(req)
	if err != nil { panic(err) }
	defer resp.Body.Close()
	out, _ := io.ReadAll(resp.Body)
	fmt.Println(string(out))
}

<?php
$payload = json_encode([
    "text1" => "今天天气不错，适合出门散步",
    "text2" => "今天天气真好，适合出门走走",
], JSON_UNESCAPED_UNICODE);

$ch = curl_init("https://v1.apizero.cn/api/text-similarity");
curl_setopt_array($ch, [
    CURLOPT_RETURNTRANSFER => true,
    CURLOPT_CUSTOMREQUEST  => "POST",
    CURLOPT_POSTFIELDS     => $payload,
    CURLOPT_HTTPHEADER     => [
        "X-Api-Key: YOUR_API_KEY",
        "Content-Type: application/json",
    ],
    CURLOPT_TIMEOUT        => 15,
]);
$body = curl_exec($ch);
curl_close($ch);

$data = json_decode($body, true);
print_r($data);

{
    "code": 0,
    "msg": "成功",
    "data": {
        "text1_length": 13,
        "text2_length": 13,
        "overall_score": 0.6471,
        "similarity_level": "moderately_similar",
        "level_name": "中度相似",
        "metrics": {
            "cosine": 0.5833,
            "jaccard": 0.4118,
            "edit_similarity": 0.6923,
            "lcs_similarity": 0.9231,
            "edit_distance": 4,
            "lcs_length": 12
        },
        "truncated": false
    },
    "request_id": "abc123def456"
}

• 1.0.0(2026-05-07)
  • 首次上线，纯本地计算无上游依赖
  • 4 种相似度指标加权融合：余弦 35% + Jaccard 25% + Edit 20% + LCS 20%
  • 修复源码 BUG：自实现 mb_levenshtein 按字符（非字节）计算编辑距离，杜绝中文 85 字以上崩溃问题
  • 修复源码 BUG：源码 levenshtein else 分支用 mb_substr(0,200) 截取后仍可能超 255 字节限制 → 已用纯字符级算法替代
  • LCS 空间优化为 O(min(m,n))，5000×5000 比对内存 < 200KB
  • 超过 500 字符自动截取并按比例还原 edit_distance（保证 P99 响应 < 100ms）
  • 新增 truncated 字段，告知前端是否启用了截取近似
  • 新增 metrics.lcs_length 字段，便于查重场景直接读取「连续相同片段长度」
  • 5 级评级：nearly_identical(≥0.9) / highly_similar(≥0.7) / moderately_similar(≥0.4) / slightly_similar(≥0.2) / different
  • 支持 application/json 请求体