geo/server/internal/tenant/app/question_metadata_classify.go

package app

import (
	"strings"
	"unicode"
	"unicode/utf8"
)

const (
	QuestionLayerL1 = "L1"
	QuestionLayerL2 = "L2"
	QuestionLayerL3 = "L3"
	QuestionLayerL4 = "L4"
	QuestionLayerL5 = "L5"

	QuestionIntentInformational = "informational"
	QuestionIntentEvaluative    = "evaluative"
	QuestionIntentDecisional    = "decisional"

	QuestionSourceManual      = "manual"
	QuestionSourceCombination = "combination"
	QuestionSourceAIDistill   = "ai_distill"
)

type ClassifiedQuestion struct {
	Text   string `json:"text"`
	Layer  string `json:"layer"`
	Intent string `json:"intent"`
}

func normalizeQuestionText(text string) string {
	trimmed := strings.TrimSpace(text)
	if trimmed == "" {
		return ""
	}
	return strings.Map(func(r rune) rune {
		if unicode.IsPunct(r) {
			return -1
		}
		return r
	}, trimmed)
}

func normalizeQuestionKey(text string) string {
	return strings.ToLower(normalizeQuestionText(text))
}

func isValidQuestionLayer(value string) bool {
	switch strings.TrimSpace(value) {
	case QuestionLayerL1, QuestionLayerL2, QuestionLayerL3, QuestionLayerL4, QuestionLayerL5:
		return true
	default:
		return false
	}
}

func isValidQuestionIntent(value string) bool {
	switch strings.TrimSpace(value) {
	case QuestionIntentInformational, QuestionIntentEvaluative, QuestionIntentDecisional:
		return true
	default:
		return false
	}
}

func isValidExternalQuestionSource(value string) bool {
	switch strings.TrimSpace(value) {
	case QuestionSourceManual, QuestionSourceCombination, QuestionSourceAIDistill:
		return true
	default:
		return false
	}
}

func inferQuestionIntent(text string) string {
	normalized := strings.ToLower(strings.TrimSpace(text))
	if containsAny(normalized, []string{"哪家好", "哪个好", "怎么选", "推荐", "哪里买", "适合谁", "多少钱", "报价", "购买"}) {
		return QuestionIntentDecisional
	}
	if containsAny(normalized, []string{"对比", "区别", "哪个更", "优劣", "测评", "排名", "值不值", "替代", "竞品"}) {
		return QuestionIntentEvaluative
	}
	return QuestionIntentInformational
}

func inferQuestionLayer(text, brandName string, competitorNames []string) string {
	normalized := strings.ToLower(strings.TrimSpace(text))
	brandHits := 0
	if containsFold(normalized, brandName) {
		brandHits++
	}
	for _, competitor := range competitorNames {
		if containsFold(normalized, competitor) {
			brandHits++
		}
	}
	if brandHits >= 2 || containsAny(normalized, []string{"对比", "区别", "哪个更", "优劣", "竞品", "替代"}) {
		return QuestionLayerL5
	}
	if brandHits == 1 {
		return QuestionLayerL1
	}
	if containsAny(normalized, []string{"北京", "上海", "广州", "深圳", "杭州", "成都", "合肥", "华东", "华南", "华北", "中小企业", "企业", "团队", "老板", "新手", "预算", "价格", "费用", "场景", "本地"}) {
		return QuestionLayerL4
	}
	if containsAny(normalized, []string{"痛点", "问题", "踩坑", "解决", "失败", "难", "风险", "怎么办"}) {
		return QuestionLayerL3
	}
	return QuestionLayerL2
}

func classifyQuestionText(text, brandName string, competitorNames []string) ClassifiedQuestion {
	trimmed := normalizeQuestionText(text)
	return ClassifiedQuestion{
		Text:   trimmed,
		Layer:  inferQuestionLayer(trimmed, brandName, competitorNames),
		Intent: inferQuestionIntent(trimmed),
	}
}

func questionLooksValid(text string) bool {
	trimmed := normalizeQuestionText(text)
	if utf8.RuneCountInString(trimmed) < 4 {
		return false
	}
	hasLetterOrNumber := false
	for _, r := range trimmed {
		if unicode.IsLetter(r) || unicode.IsNumber(r) {
			hasLetterOrNumber = true
			break
		}
	}
	if !hasLetterOrNumber {
		return false
	}
	return containsAny(strings.ToLower(trimmed), []string{"?", "？", "什么", "如何", "怎么", "哪", "为什么", "是否", "能不能", "适合", "区别", "对比", "推荐"})
}

func containsAny(value string, needles []string) bool {
	for _, needle := range needles {
		if strings.Contains(value, strings.ToLower(strings.TrimSpace(needle))) {
			return true
		}
	}
	return false
}

func containsFold(value, needle string) bool {
	needle = strings.ToLower(strings.TrimSpace(needle))
	return needle != "" && strings.Contains(value, needle)
}