ElasticSearch學習筆記-相關度得分記錄

最近想要修改調整一下ElasticSearch裏面Doc的Score，因而在ES官網查閱了一下，相關的介紹和說明仍是很是詳細的，能作的修改調整也仍是比較多的，須要根據具體的情形去選擇相應的方式作合適的調整修改，這裏作個簡單的記錄，以便後續使用方便。

相關解釋描述能夠參考鏈接：

https://www.elastic.co/guide/cn/elasticsearch/guide/current/controlling-relevance.html

https://www.elastic.co/guide/cn/elasticsearch/guide/current/function-score-query.html

https://www.elastic.co/guide/cn/elasticsearch/guide/current/boosting-by-popularity.html

https://www.elastic.co/guide/cn/elasticsearch/guide/current/function-score-filters.html

https://www.elastic.co/guide/cn/elasticsearch/guide/current/random-scoring.html

https://www.elastic.co/guide/cn/elasticsearch/guide/current/decay-functions.html

https://www.elastic.co/guide/cn/elasticsearch/guide/current/script-score.html

 

如下是API操做的部分記錄：

 

ScoreFunctionBuilder scoreFunctionBuilder = 
ScoreFunctionBuilders.fieldValueFactorFunction(fieldName)
	.factor(boostFactor).modifier(modifier).missing(missing).setWeight(weight);
線性衰減函數  一旦直線與橫軸 0 相交，全部其餘值的評分都是 0.0
ScoreFunctionBuilders.linearDecayFunction(fieldName, origin, scale)
	.setDecay(decay).setOffset(offset).setWeight(weight)
指數衰減函數  先劇烈衰減而後變緩 
ScoreFunctionBuilders.exponentialDecayFunction(fieldName, origin, scale)
	.setDecay(decay).setOffset(offset).setWeight(weight)
高斯衰減函數  高斯函數是鐘形的——它的衰減速率是先緩慢，而後變快，最後又放緩。
ScoreFunctionBuilders.gaussDecayFunction(fieldName, origin, scale)
	.setDecay(decay).setOffset(offset).setWeight(weight)
origin  中心點或字段可能的最佳值，落在原點 origin上的文檔評分 _score 爲滿分 1.0 。
scale  衰減率，即一個文檔從原點 origin下落時，評分 _score改變的速度。
decay  從原點 origin衰減到 scale所得的評分 _score，默認值爲 0.5 。
offset  以原點 origin爲中心點，爲其設置一個非零的偏移量 offset覆蓋一個範圍，而不僅是單個原點。
	在範圍 -offset <= origin <= +offset內的全部評分 _score 都是 1.0 。
隨機評分
ScoreFunctionBuilders.randomFunction(seed)
權重因素
ScoreFunctionBuilders.weightFactorFunction(weight)

 

腳本評分

1. String timeField = getTypeTimeField(type);

2. if (StringUtils.isBlank(timeField)) {

3. searchRequestBuilder.setQuery(buildBoolQuery(params.keywords(), attributes));

4. } else {

5. String inlineScript = ElasticScriptUtils.scriptWithScoreAndTime(timeField, System.currentTimeMillis());

6. Map<String, Object> sparams = new HashMap<>();

7. Script script = new Script(inlineScript, ScriptType.INLINE, "groovy", sparams);

8. ScoreFunctionBuilder scoreFunctionBuilder = ScoreFunctionBuilders.scriptFunction(script);

9. searchRequestBuilder.setQuery(QueryBuilders.functionScoreQuery(

10. buildBoolQuery(params.keywords(), attributes), scoreFunctionBuilder));

11. }

 

1. public class ElasticScriptUtils {

2.  

3. public static String scriptWithScoreAndTimeV1(String field, long currentTime) {

4. String script = ""

5. + "field = (null==_source." + field + "?\"1970-01-01 12:00:00\":source." + field + ");"

6. + "format = \"\";"

7. + "ds = field.trim().split(\":\");"

8. + "if (ds.length == 3) {"

9. + " format = \"yyyy-MM-dd HH:mm:ss\";"

10. + "} else if (ds.length == 2) {"

11. + " format = \"yyyy-MM-dd HH:mm\";"

12. + "} else if (ds.length == 1) {"

13. + " ds = d.trim().split(\" \");"

14. + " if (ds.length == 2) {"

15. + " format = \"yyyy-MM-dd HH\";"

16. + " } else if (ds.length == 1) {"

17. + " ds = d.trim().split(\"-\");"

18. + " if (ds.length == 3) {"

19. + " format = \"yyyy-MM-dd\";"

20. + " } else if (ds.length == 2) {"

21. + " format = \"yyyy-MM\";"

22. + " } else if (ds.length == 1) {"

23. + " format = \"yyyy\";"

24. + " }"

25. + " }"

26. + "};"

27. + "parse_date = Date.parse(format, field).getTime();"

28. + "return _score.doubleValue() + (parse_date / " + currentTime + ");";

29. return script;

30. }

31.  

32. public static String scriptWithScoreAndTime(String fields, long currentTime) {

33. String script = ""

34. + "temp_fields = \"" + fields + "\".trim().split(\",\");"

35. + "target_field = temp_fields[0];"

36. + "temp_field = _source.target_field;"

37. + "for (i in 2.. temp_fields.length) {"

38. + " if (null != temp_field) break;"

39. + " target_field = temp_fields[i-1];"

40. + " temp_field = _source.target_field;"

41. + "};"

42. + "field = (null==temp_field ? \"1970-01-01 12:00:00\" : temp_field);"

43. + "format = \"\";"

44. + "ds = field.trim().split(\":\");"

45. + "if (ds.length == 3) {"

46. + " format = \"yyyy-MM-dd HH:mm:ss\";"

47. + "} else if (ds.length == 2) {"

48. + " format = \"yyyy-MM-dd HH:mm\";"

49. + "} else if (ds.length == 1) {"

50. + " ds = d.trim().split(\" \");"

51. + " if (ds.length == 2) {"

52. + " format = \"yyyy-MM-dd HH\";"

53. + " } else if (ds.length == 1) {"

54. + " ds = d.trim().split(\"-\");"

55. + " if (ds.length == 3) {"

56. + " format = \"yyyy-MM-dd\";"

57. + " } else if (ds.length == 2) {"

58. + " format = \"yyyy-MM\";"

59. + " } else if (ds.length == 1) {"

60. + " format = \"yyyy\";"

61. + " }"

62. + " }"

63. + "};"

64. + "parse_date = Date.parse(format, field).getTime();"

65. /**

66. + "println _score.doubleValue(); println (parse_date / " + currentTime + ");"

67. **/

68. + "return _score.doubleValue() + (parse_date / " + currentTime + ");";

69. return script;

70. }

71.  

 

https://www.elastic.co/guide/cn/elasticsearch/guide/current/decay-functions.html