Merge branch 'feature/luojunhui/20260803-publish-times-strategy-improve' of Server/long-article-recommend into master

long-article-recommend-service/src/main/java/com/tzld/longarticle/recommend/server/common/enums/recommend/RankStrategyEnum.java

@@ -27,6 +27,7 @@ public enum RankStrategyEnum {
     ArticleRankV19("ArticleRankV19", "ArticleRankV19", "rankV19Strategy"),
     ArticleRankV20("ArticleRankV20", "ArticleRankV20", "rankV20Strategy"),
     ArticleRankV21("ArticleRankV21", "ArticleRankV21", "rankV21Strategy"),
+    ArticleRankV22("ArticleRankV22", "ArticleRankV22", "rankV22Strategy"),
 
     HIS_JUMP_STRATEGY("ArticleRankHisJump", "历史表现跳过相似度策略", "hisJumpRankStrategy"),
     INFINITE_STRATEGY("ArticleRankInfinite", "无限发表", "infiniteRankStrategy"),

long-article-recommend-service/src/main/java/com/tzld/longarticle/recommend/server/common/enums/recommend/ScoreStrategyEnum.java

@@ -24,6 +24,7 @@ public enum ScoreStrategyEnum {
     CRAWLER_DAYS_DECREASE_STRATEGY("CrawlerDaysDecreaseStrategy"),
     CRAWLER_DAYS_DECREASE_V2_STRATEGY("CrawlerDaysDecreaseV2Strategy"),
     I2I_RECOMMEND_STRATEGY("I2IRecommendStrategy"),
+    CROSS_ACCOUNT_MUTEX("CrossAccountMutexStrategy"),
     ;
 
     private final String value;

long-article-recommend-service/src/main/java/com/tzld/longarticle/recommend/server/service/recommend/rank/strategy/RankV22Strategy.java

@@ -0,0 +1,219 @@
+package com.tzld.longarticle.recommend.server.service.recommend.rank.strategy;
+
+
+import com.ctrip.framework.apollo.spring.annotation.ApolloJsonValue;
+import com.tzld.longarticle.recommend.server.common.enums.aigc.PublishPlanInputSourceTypesEnum;
+import com.tzld.longarticle.recommend.server.common.enums.recommend.ScoreStrategyEnum;
+import com.tzld.longarticle.recommend.server.model.dto.Content;
+import com.tzld.longarticle.recommend.server.model.entity.crawler.Article;
+import com.tzld.longarticle.recommend.server.repository.crawler.ArticleRepository;
+import com.tzld.longarticle.recommend.server.service.recommend.config.AccountContentPoolConfigService;
+import com.tzld.longarticle.recommend.server.service.recommend.config.StrategyIndexScoreWeightService;
+import com.tzld.longarticle.recommend.server.service.recommend.rank.*;
+import com.tzld.longarticle.recommend.server.service.recommend.score.AccountIndexReplacePoolConfig;
+import com.tzld.longarticle.recommend.server.service.recommend.score.ScoreResult;
+import com.tzld.longarticle.recommend.server.service.recommend.score.ScoreService;
+import com.tzld.longarticle.recommend.server.util.CommonCollectionUtils;
+import lombok.extern.slf4j.Slf4j;
+import org.apache.commons.collections4.CollectionUtils;
+import org.apache.commons.lang3.RandomUtils;
+import org.springframework.beans.factory.annotation.Autowired;
+import org.springframework.beans.factory.annotation.Value;
+import org.springframework.stereotype.Service;
+
+import java.util.*;
+import java.util.stream.Collectors;
+
+@Service
+@Slf4j
+public class RankV22Strategy implements RankStrategy {
+
+    @Autowired
+    private ScoreService scoreService;
+    @Autowired
+    private RankService rankService;
+    @Autowired
+    private AccountContentPoolConfigService accountContentPoolConfigService;
+    @Autowired
+    private ArticleRepository articleRepository;
+    @Autowired
+    private StrategyIndexScoreWeightService weightService;
+
+    @ApolloJsonValue("${touliu.account.ghIds:[\"gh_93e00e187787\", \"gh_ac43e43b253b\", \"gh_68e7fdc09fe4\",\"gh_77f36c109fb1\", \"gh_b181786a6c8c\", \"gh_1ee2e1b39ccf\"]}")
+    private List<String> touliuAccountGhIds;
+    @Value("${topProducePlanId:}")
+    private String topProducePlanId;
+    @Value("${hisFissionOpenRateMissingBaseline:0.32}")
+    private double hisFissionOpenRateMissingBaseline;
+
+    public RankResult rank(RankParam param) {
+        List<Content> result = new ArrayList<>();
+
+        ScoreResult scoreResult = scoreService.score(RankStrategy.convertToScoreParam(param));
+
+        Map<String, Map<String, Double>> scoreMap = scoreResult.getScoreMap();
+        String[] contentPools = accountContentPoolConfigService.getContentPools(param.getAccountName());
+        Map<Integer, AccountIndexReplacePoolConfig> indexReplacePoolConfigMap = accountContentPoolConfigService.getContentReplacePools(param.getAccountName());
+
+        List<RankItem> items = CommonCollectionUtils.toList(param.getContents(), c -> {
+            RankItem item = new RankItem();
+            item.setContent(c);
+            c.setScoreMap(scoreMap.get(c.getId()));
+            item.setScoreMap(scoreMap.get(c.getId()));
+            double score;
+            int index = weightService.getIndex(item.getContent().getContentPoolType(), contentPools);
+            if (contentPools[0].equals(item.getContent().getContentPoolType())) {
+                score = item.getScore(ScoreStrategyEnum.I2I_RECOMMEND_STRATEGY.value())
+                        * weightService.getWeight(param.getStrategy(), param.getGhId(), index,
+                        ScoreStrategyEnum.I2I_RECOMMEND_STRATEGY.value())
+                        + item.getScore(ScoreStrategyEnum.CATEGORY.value())
+                        * weightService.getWeight(param.getStrategy(), param.getGhId(), index,
+                        ScoreStrategyEnum.CATEGORY.value())
+                        + item.getScore(ScoreStrategyEnum.ACCOUNT_USER_CATEGORY.value())
+                        * weightService.getWeight(param.getStrategy(), param.getGhId(), index,
+                        ScoreStrategyEnum.ACCOUNT_USER_CATEGORY.value())
+                        + item.getScore(ScoreStrategyEnum.HIS_FISSION_OPEN_RATE.value())
+                        * weightService.getWeight(param.getStrategy(), param.getGhId(), index,
+                        ScoreStrategyEnum.HIS_FISSION_OPEN_RATE.value())
+                        + item.getScore(ScoreStrategyEnum.FLOW_CTL_DECREASE.value())
+                        + item.getScore(ScoreStrategyEnum.CRAWLER_DAYS_DECREASE_V2_STRATEGY.value())
+                        * weightService.getWeight(param.getStrategy(), param.getGhId(), index,
+                        ScoreStrategyEnum.CRAWLER_DAYS_DECREASE_V2_STRATEGY.value())
+                        + item.getScore(ScoreStrategyEnum.CROSS_ACCOUNT_MUTEX.value());
+                if (item.getScore(ScoreStrategyEnum.PUBLISH_TIMES.value()) >= 0) {
+                    score += item.getScore(ScoreStrategyEnum.VIEW_COUNT_RATE.value())
+                            * weightService.getWeight(param.getStrategy(), param.getGhId(), index,
+                            ScoreStrategyEnum.VIEW_COUNT_RATE.value());
+                }
+                if (item.getScore(ScoreStrategyEnum.HIS_FISSION_OPEN_RATE.value()) == 0) {
+                    score += hisFissionOpenRateMissingBaseline * weightService.getWeight(param.getStrategy(), param.getGhId(), index,
+                            ScoreStrategyEnum.HIS_FISSION_OPEN_RATE.value());
+                }
+            } else if (contentPools[1].equals(item.getContent().getContentPoolType())) {
+                score = item.getScore(ScoreStrategyEnum.SIMILARITY.value())
+                        * weightService.getWeight(param.getStrategy(), param.getGhId(), index,
+                        ScoreStrategyEnum.SIMILARITY.value())
+                        + item.getScore(ScoreStrategyEnum.CATEGORY.value())
+                        * weightService.getWeight(param.getStrategy(), param.getGhId(), index,
+                        ScoreStrategyEnum.CATEGORY.value())
+                        + item.getScore(ScoreStrategyEnum.ACCOUNT_USER_CATEGORY.value())
+                        * weightService.getWeight(param.getStrategy(), param.getGhId(), index,
+                        ScoreStrategyEnum.ACCOUNT_USER_CATEGORY.value())
+                        + item.getScore(ScoreStrategyEnum.HIS_FISSION_OPEN_RATE.value())
+                        * weightService.getWeight(param.getStrategy(), param.getGhId(), index,
+                        ScoreStrategyEnum.HIS_FISSION_OPEN_RATE.value())
+                        + item.getScore(ScoreStrategyEnum.FLOW_CTL_DECREASE.value())
+                        + item.getScore(ScoreStrategyEnum.CRAWLER_DAYS_DECREASE_STRATEGY.value())
+                        * weightService.getWeight(param.getStrategy(), param.getGhId(), index,
+                        ScoreStrategyEnum.CRAWLER_DAYS_DECREASE_STRATEGY.value());
+                if (item.getScore(ScoreStrategyEnum.PUBLISH_TIMES.value()) >= 0) {
+                    score += item.getScore(ScoreStrategyEnum.VIEW_COUNT_RATE.value())
+                            * weightService.getWeight(param.getStrategy(), param.getGhId(), index,
+                            ScoreStrategyEnum.VIEW_COUNT_RATE.value());
+                }
+                if (item.getScore(ScoreStrategyEnum.HIS_FISSION_OPEN_RATE.value()) == 0) {
+                    score += hisFissionOpenRateMissingBaseline * weightService.getWeight(param.getStrategy(), param.getGhId(), index,
+                            ScoreStrategyEnum.HIS_FISSION_OPEN_RATE.value());
+                }
+            } else {
+                score = item.getScore(ScoreStrategyEnum.SIMILARITY.value())
+                        * weightService.getWeight(param.getStrategy(), param.getGhId(), index,
+                        ScoreStrategyEnum.SIMILARITY.value())
+                        + item.getScore(ScoreStrategyEnum.CATEGORY.value())
+                        * weightService.getWeight(param.getStrategy(), param.getGhId(), index,
+                        ScoreStrategyEnum.CATEGORY.value())
+                        + item.getScore(ScoreStrategyEnum.ACCOUNT_USER_CATEGORY.value())
+                        * weightService.getWeight(param.getStrategy(), param.getGhId(), index,
+                        ScoreStrategyEnum.ACCOUNT_USER_CATEGORY.value())
+                        + item.getScore(ScoreStrategyEnum.ACCOUNT_PRE_DISTRIBUTE.value())
+                        + item.getScore(ScoreStrategyEnum.PUBLISH_TIMES.value())
+                        + item.getScore(ScoreStrategyEnum.CRAWLER_DAYS_DECREASE_STRATEGY.value())
+                        + item.getScore(ScoreStrategyEnum.FLOW_CTL_DECREASE.value());
+            }
+            c.setScore(score);
+            item.setScore(score);
+            return item;
+        });
+        // 相似度评分为0 报警返回
+        List<Article> hisPublishFirstArticleList = articleRepository.getByGhIdAndItemIndexAndTypeEqualsAndStatusEquals(
+                param.getGhId(), 1, param.getType(), 1);
+        if (RankStrategy.SimilarityScoreZero(items, param, hisPublishFirstArticleList)) {
+            return new RankResult(result);
+        }
+        // 安全分降权
+        //RankService.safeScoreDecrease(items);
+
+        // 1 排序
+        Collections.sort(items, (o1, o2) -> -Double.compare(o1.getScore(), o2.getScore()));
+        // 2 相似去重
+        List<Content> contents = CommonCollectionUtils.toList(items, RankItem::getContent);
+
+        // 异步保存排序内容分数（保存所有排序后的内容）
+        RankStrategy.asyncSaveRankContentScore(param, items);
+
+        // 3 文章按照内容池分组
+        Map<String, List<Content>> contentMap = new HashMap<>();
+        for (Content c : contents) {
+            List<Content> data = contentMap.computeIfAbsent(c.getContentPoolType(), k -> new ArrayList<>());
+            data.add(c);
+        }
+        // 4 选文章
+        String[] publishPool = Arrays.copyOf(contentPools, contentPools.length);
+
+        // 头
+        List<Content> pool1 = contentMap.get(contentPools[0]);
+        if (CollectionUtils.isNotEmpty(pool1)) {
+            pool1 = RankService.contentSourceTypeFilter(param.getStrategy(), pool1, 1);
+        }
+        RankService.printSortLog(param.getStrategy(), param.getAccountName(), "头条", pool1);
+        if (CollectionUtils.isNotEmpty(pool1)) {
+            if (topProducePlanId.equals(pool1.get(0).getProducePlanId())) {
+                int i = RandomUtils.nextInt(0, 2);
+                if (i == 0) {
+                    for (Content content : pool1) {
+                        if (!topProducePlanId.equals(content.getProducePlanId())) {
+                            result.add(content);
+                            break;
+                        }
+                    }
+                }
+            }
+            if (CollectionUtils.isEmpty(result)) {
+                result.add(pool1.get(0));
+            }
+        } else {
+            RankStrategy.sendFeishuFirstPoolEmpty(param, contentPools[0]);
+            return new RankResult(result);
+        }
+
+        // 次
+        RankService.commonAddSecondContent(param, result, publishPool, contentPools, contentMap,
+                indexReplacePoolConfigMap, param.getStrategy());
+
+        // 3-8
+        // RankService.commonAdd38Content(param, result, contentPools, contentMap, param.getStrategy());
+        List<Content> pool = contentMap.get(contentPools[2]);
+        RankService.printSortLog(param.getStrategy(), param.getAccountName(), "3-8", pool);
+        if (CollectionUtils.isNotEmpty(pool)) {
+            Integer videoSourceType = PublishPlanInputSourceTypesEnum.longArticleVideoPoolSource.getVal();
+            Queue<Content> videoPoolQueue = pool.stream().filter(o -> Objects.equals(o.getSourceType(), videoSourceType))
+                    .collect(Collectors.toCollection(LinkedList::new));
+            Queue<Content> otherPoolQueue = pool.stream().filter(o -> !Objects.equals(o.getSourceType(), videoSourceType))
+                    .collect(Collectors.toCollection(LinkedList::new));
+            for (int i = 3; i < param.getSize() + 1; i++) {
+                Integer sourceType = RankService.getStrategyPoolSourceType(param.getStrategy(), i);
+                if (Objects.equals(sourceType, videoSourceType) && !videoPoolQueue.isEmpty()) {
+                    result.add(videoPoolQueue.poll());
+                } else if (!otherPoolQueue.isEmpty()) {
+                    result.add(otherPoolQueue.poll());
+                }
+            }
+        }
+
+        rankService.checkPublishContentStatus(result, contentMap, publishPool);
+        RankStrategy.deduplication(result, contentMap, publishPool);
+
+        return new RankResult(result);
+    }
+
+}

long-article-recommend-service/src/main/java/com/tzld/longarticle/recommend/server/service/recommend/score/ScoreService.java

@@ -148,6 +148,7 @@ public class ScoreService implements ApplicationContextAware {
                 || StringUtils.equals(param.getStrategy(), RankStrategyEnum.ArticleRankV19.getStrategy())
                 || StringUtils.equals(param.getStrategy(), RankStrategyEnum.ArticleRankV20.getStrategy())
                 || StringUtils.equals(param.getStrategy(), RankStrategyEnum.ArticleRankV21.getStrategy())
+                || StringUtils.equals(param.getStrategy(), RankStrategyEnum.ArticleRankV22.getStrategy())
                 || StringUtils.equals(param.getStrategy(), RankStrategyEnum.INFINITE_STRATEGY.getStrategy())) {
             strategies.add(strategyMap.get(ScoreStrategyEnum.CATEGORY.value()));
             //strategies.add(strategyMap.get(ScoreStrategyEnum.DEDUP_CATEGORY.value()));
@@ -168,9 +169,13 @@ public class ScoreService implements ApplicationContextAware {
         }
         if (StringUtils.equals(param.getStrategy(), RankStrategyEnum.ArticleRankV20.getStrategy())
                 || StringUtils.equals(param.getStrategy(), RankStrategyEnum.ArticleRankV21.getStrategy())
+                || StringUtils.equals(param.getStrategy(), RankStrategyEnum.ArticleRankV22.getStrategy())
                 || StringUtils.equals(param.getStrategy(), RankStrategyEnum.INFINITE_STRATEGY.getStrategy())) {
             strategies.add(strategyMap.get(ScoreStrategyEnum.I2I_RECOMMEND_STRATEGY.value()));
         }
+        if (StringUtils.equals(param.getStrategy(), RankStrategyEnum.ArticleRankV22.getStrategy())) {
+            strategies.add(strategyMap.get(ScoreStrategyEnum.CROSS_ACCOUNT_MUTEX.value()));
+        }
 
         return strategies;
     }

long-article-recommend-service/src/main/java/com/tzld/longarticle/recommend/server/service/recommend/score/strategy/CrossAccountMutexStrategy.java

@@ -0,0 +1,74 @@
+package com.tzld.longarticle.recommend.server.service.recommend.score.strategy;
+
+import com.ctrip.framework.apollo.spring.annotation.ApolloJsonValue;
+import com.tzld.longarticle.recommend.server.model.dto.Content;
+import com.tzld.longarticle.recommend.server.model.entity.crawler.PublishSortLog;
+import com.tzld.longarticle.recommend.server.repository.crawler.PublishSortLogRepository;
+import com.tzld.longarticle.recommend.server.service.recommend.score.Score;
+import com.tzld.longarticle.recommend.server.service.recommend.score.ScoreParam;
+import com.tzld.longarticle.recommend.server.service.recommend.score.ScoreStrategy;
+import com.tzld.longarticle.recommend.server.util.DateUtils;
+import lombok.extern.slf4j.Slf4j;
+import org.springframework.beans.factory.annotation.Autowired;
+import org.springframework.beans.factory.annotation.Value;
+import org.springframework.stereotype.Component;
+import org.springframework.util.CollectionUtils;
+
+import java.util.*;
+import java.util.stream.Collectors;
+
+/**
+ * 跨账号头条互斥策略：对于互斥账号集合，若某标题今天已被集合中其他账号在头条(index=1)发布过，
+ * 则当前账号对该标题的内容给与降权（负分），避免不同账号在头条重复发布相同标题的文章。
+ */
+@Component
+@Slf4j
+public class CrossAccountMutexStrategy implements ScoreStrategy {
+
+    @Autowired
+    private PublishSortLogRepository publishSortLogRepository;
+
+    @ApolloJsonValue("${crossAccountMutexGhIds:[]}")
+    private List<String> crossAccountMutexGhIds;
+
+    @Value("${crossAccountMutexWeight:-1000}")
+    private double crossAccountMutexWeight;
+
+    @Override
+    public List<Score> score(ScoreParam param) {
+        long start = System.currentTimeMillis();
+        List<Score> scores = new ArrayList<>();
+        if (CollectionUtils.isEmpty(param.getContents())) {
+            return scores;
+        }
+        // 当前账号不在互斥集合中，跳过
+        if (CollectionUtils.isEmpty(crossAccountMutexGhIds)
+                || !crossAccountMutexGhIds.contains(param.getGhId())) {
+            return scores;
+        }
+        // 查询当天头条已发布记录，收集其他互斥账号发布过的标题
+        String dateStr = DateUtils.getCurrentDateStr("yyyyMMdd");
+        List<PublishSortLog> todayPublished = publishSortLogRepository.getByGhIdInAndDateStrIn(
+                crossAccountMutexGhIds, Collections.singletonList(dateStr));
+        Set<String> publishedTitles = todayPublished.stream()
+                .filter(o -> Objects.equals(o.getIndex(), 1))
+                .filter(o -> !Objects.equals(o.getGhId(), param.getGhId()))
+                .filter(o -> Objects.nonNull(o.getTitle()))
+                .map(PublishSortLog::getTitle)
+                .collect(Collectors.toSet());
+
+        for (Content content : param.getContents()) {
+            Score score = new Score();
+            score.setStrategy(this);
+            score.setContentId(content.getId());
+            if (publishedTitles.contains(content.getTitle())) {
+                score.setScore(crossAccountMutexWeight);
+            } else {
+                score.setScore(0.0);
+            }
+            scores.add(score);
+        }
+        log.info("CrossAccountMutexStrategy cost:{}", System.currentTimeMillis() - start);
+        return scores;
+    }
+}