An Introduction to NLP4L
3 likes3,754 views
The document introduces NLP4L, a natural language processing tool for Apache Lucene. NLP4L aims to improve search experiences by utilizing NLP/ML technologies to generate models, dictionaries and indexes from text corpora. It provides a framework that allows plug-ins for various NLP tasks like transliteration, named entity extraction and document classification. The framework also includes interfaces for users to examine and modify generated outputs.
![Make them Structured
I
D
article person org loc
1
David Cameron says he has a mandate to pursue EU reform following
the Conservatives' general election victory. The Prime Minister will be
hoping his majority government will give him extra leverage in Brussels.
David
Cameron
EU
Bruss
els
2
He wants to renegotiate the terms of the UK's membership ahead of a
referendum by the end of 2017. He has said he will campaign for
Britain to remain in the EU if he gets the reforms he wants.
EU
UK
Britai
n
NEE[1] extracts interesting words.
[1] Named Entity Extraction
22](https://image.slidesharecdn.com/uploadnlp4l-intro-meetup-150824004024-lva1-app6892/85/An-Introduction-to-NLP4L-22-320.jpg)
An Introduction to NLP4L
- 1. An Introduction to NLP4L Natural Language Processing tool for Apache Lucene Koji Sekiguchi @kojisays Founder & CEO, RONDHUIT
- 2. My contributions • CharFilter framework & MappingCharFilter • FastVectorHighlighter 2
- 3. Agenda • What s NLP4L? • How NLP improves search experience • Calculate probabilities using ShingleFilter • Transliteration (Application for HMM) • NLP4L Framework (coming soon) 3
- 4. Agenda • What s NLP4L? • How NLP improves search experience • Calculate probabilities using ShingleFilter • Transliteration (Application for HMM) • NLP4L Framework (coming soon) 4
- 6. What s NLP4L? • GOAL • Improve Lucene users search experience • FEATURES • Use of Lucene index as a Corpus Database • Lucene API Front-end written in Scala • NLP4L provides • Preprocessors for existing ML tools • Provision of ML algorithms and Applications (e.g. Transliteration) 6
- 7. What s NLP4L? • GOAL • Improve Lucene users search experience • FEATURES • Use of Lucene index as a Corpus Database • Lucene API Front-end written in Scala • NLP4L provides • Preprocessors for existing ML tools • Provision of ML algorithms and Applications (e.g. Transliteration) 7
- 8. What s NLP4L? • GOAL • Improve Lucene users search experience • FEATURES • Use of Lucene index as a Corpus Database • Lucene API Front-end written in Scala • NLP4L provides • Preprocessors for existing ML tools • Provision of ML algorithms and Applications (e.g. Transliteration) 8
- 9. Agenda • What s NLP4L? • How NLP improves search experience • Calculate probabilities using ShingleFilter • Transliteration (Application for HMM) • NLP4L Framework (coming soon) 9
- 10. Evaluation Measures targetresult tpfp fn tn precision = tp / (tp + fp) recall = tp / (tp + fn) 10
- 11. Recall ,Precision tpfp fn tn precision = tp / (tp + fp) recall = tp / (tp + fn) 11
- 12. targetresult tpfp fn tn precision = tp / (tp + fp) recall = tp / (tp + fn) Recall ,Precision 12
- 13. Solution n-gram, synonym dictionary, etc. facet (filter query) Ranking Tuning recall precision recall , precision 13
- 14. Solution n-gram, synonym dictionary, etc. facet (filter query) Ranking Tuning recall precision recall , precision 14
- 15. Solution n-gram, synonym dictionary, etc. e.g. Transliteration facet (filter query) recall precision recall , precision Ranking Tuning 15
- 16. Solution n-gram, synonym dictionary, etc. e.g. Transliteration facet (filter query) e.g. Named Entity Extraction recall precision recall , precision Ranking Tuning 16
- 18. filter by Gender=Men s targetresult gradual precision improvement 18
- 19. targetresult filter by Gender=Men s filter by Price=100-150 gradual precision improvement 19
- 20. Structured Documents ID product price gender 1 CURREN New Men s Date Stainless Steel Military Sport Quartz Wrist Watch 8.92 Men s 2 Suiksilver The Gamer Watch 87.99 Men s 20
- 21. Unstructured Documents ID article 1 David Cameron says he has a mandate to pursue EU reform following the Conservatives' general election victory. The Prime Minister will be hoping his majority government will give him extra leverage in Brussels. 2 He wants to renegotiate the terms of the UK's membership ahead of a referendum by the end of 2017. He has said he will campaign for Britain to remain in the EU if he gets the reforms he wants. 21
- 22. Make them Structured I D article person org loc 1 David Cameron says he has a mandate to pursue EU reform following the Conservatives' general election victory. The Prime Minister will be hoping his majority government will give him extra leverage in Brussels. David Cameron EU Bruss els 2 He wants to renegotiate the terms of the UK's membership ahead of a referendum by the end of 2017. He has said he will campaign for Britain to remain in the EU if he gets the reforms he wants. EU UK Britai n NEE[1] extracts interesting words. [1] Named Entity Extraction 22
- 23. Manual Tagging using brat 23
- 24. Agenda • What s NLP4L? • How NLP improves search experience • Calculate probabilities using ShingleFilter • Transliteration (Application for HMM) • NLP4L Framework (coming soon) 24
- 25. Language Model • LM represents the fluency of language • N-gram model is the LM which is most widely used • Calculation example for 2-gram totalTermFreq(”word2g”,”an apple”) totalTermFreq(”word”,”an”) 25
- 26. Alice/NNP ate/VB an/AT apple/NNP ./. Mike/NNP likes/VB an/AT orange/NNP ./. An/AT apple/NNP is/VB red/JJ ./. NNP Proper noun, singular VB Verb AT Article JJ Adjective . period Part-of-Speech Tagging Our Corpus for training 26
- 28. Hidden Markov Model Series of Words 28
- 29. Hidden Markov Model Series of Part-of-Speech 29
- 32. HMM state diagram NNP 0.667 VB 0.0 . 0.0 JJ 0.0 AT 0.333 1.0 1.0 0.4 0.6 0.6670.333 alice 0.2 apple 0.4 mike 0.2 orange 0.2 ate 0.333 is 0.333 likes 0.333 an 1.0 red 1.0 . 1.0 32
- 33. Agenda • What s NLP4L? • How NLP improves search experience • Calculate probabilities using ShingleFilter • Transliteration (Application for HMM) • NLP4L Framework (coming soon) 33
- 34. Transliteration Transliteration is a process of transcribing letters or words from one alphabet to another one to facilitate comprehension and pronunciation for non-native speakers. computer コンピューター server サーバー internet インターネット mouse マウス information インフォメーション examples of transliteration from English to Japanese 34
- 35. It helps improve recall you search English mouse 35
- 36. It helps improve recall but you got マウス (=mouse) highlighted in Japanese 36
- 37. Training data in NLP4L アaカcaデdeミーmy アaクcセceンnトt アaクcセceスss アaクcシciデdeンnトt アaクcロroバッbaトt アaクcショtioンn アaダdaプpターter アaフfリriカca エaアirバbuスs アaラlaスsカka アaルlコーcohoルl アaレlleルrギーgy train_data/alpha_katakana.txt train_data/alpha_katakana_aligned.txt academy,アカデミー accent,アクセント access,アクセス accident,アクシデント acrobat,アクロバット action,アクション adapter,アダプター africa,アフリカ airbus,エアバス alaska,アラスカ alcohol,アルコール allergy,アレルギー 37
- 38. Demo: Transliteration Input Prediction Right Answer アルゴリズム algorism algorithm プログラム program (OK) ケミカル chaemmical chemical ダイニング dining (OK) コミッター committer (OK) エントリー entree entry nlp4l> :load examples/trans_katakana_alpha.scala 38
- 39. Gathering loan words ① crawl gathering Katakana-Alphabet string pairs アルゴリズム, algorithm Transliteration アルゴリズム algorism calculate edit distance synonyms.txt store pair of strings if edit distance is small enough ② ③ ④ ⑤ ⑥ 39
- 40. Gathering loan words ① crawl gathering Katakana-Alphabet string pairs アルゴリズム, algorithm Transliteration アルゴリズム algorism calculate edit distance synonyms.txt store pair of strings if edit distance is small enough ② ③ ④ ⑤ ⑥ Got 1,800+ records of synonym knowledge from jawiki 40
- 41. Agenda • What s NLP4L? • How NLP improves search experience • Calculate probabilities using ShingleFilter • Transliteration (Application for HMM) • NLP4L Framework (coming soon) 41
- 42. NLP4L Framework • A framework that improves search experience (for mainly Lucene-based search system). Pluggable. • Reference implementation of plug-ins and corpora provided. • Uses NLP/ML technologies to output models, dictionaries and indexes. • Since NLP/ML are not perfect, an interface that enables users to personally examine output dictionaries is provided as well. 42
- 43. NLP4L Framework • A framework that improves search experience (for mainly Lucene-based search system). Pluggable. • Reference implementation of plug-ins and corpora provided. • Uses NLP/ML technologies to output models, dictionaries and indexes. • Since NLP/ML are not perfect, an interface that enables users to personally examine output dictionaries is provided as well. 43
- 44. NLP4L Framework • A framework that improves search experience (for mainly Lucene-based search system). Pluggable. • Reference implementation of plug-ins and corpora provided. • Uses NLP/ML technologies to output models, dictionaries and indexes. • Since NLP/ML are not perfect, an interface that enables users to personally examine output dictionaries is provided as well. 44
- 45. NLP4L Framework • A framework that improves search experience (for mainly Lucene-based search system). Pluggable. • Reference implementation of plug-ins and corpora provided. • Uses NLP/ML technologies to output models, dictionaries and indexes. • Since NLP/ML are not perfect, an interface that enables users to personally examine output dictionaries is provided as well. 45
- 46. Solr ES Mahout Spark Data Source ・Corpus (Text data, Lucene index) ・Query Log ・Access Log Dictionaries ・Suggestion (auto complete) ・Did you mean? ・synonyms.txt ・userdic.txt ・keyword attachment maintenance Model files Tagged Corpus Document Vectors ・TermExtractor ・Transliteration ・NEE ・Classification ・Document Vectors ・Language Detection ・Learning to Rank ・Personalized Search 46
- 47. Keyword Attachment • Keyword attachment is a general format that enables the following functions. • Learning to Rank • Personalized Search • Named Entity Extraction • Document Classification Lucene doc Lucene doc keyword ↑ Increase boost 47
- 48. Before Learning to Rank targetresult 1 2 3 … 50 100 500 … 48
- 49. After Learning to Rank targetresult 1 2 3 … 50 100 500 … 49
- 50. Learning to Rank • Program learns, from access log and other sources, that the score of document d for a query q should be larger than the normal score(q,d) Lucene doc d q, q, … https://en.wikipedia.org/wiki/Learning_to_rank 50
- 51. Personalized Search targetresult 1 2 3 … 50 100 500 … q=apple computer … 51
- 52. Personalized Search target result 50 100 500 … 1 2 3 … q=applefruit … 52
- 53. Personalized Search • Program learns, from access log and other sources, that the score of document d for a query q by user u should be larger than the normal score(q,d) • Since you cannot specify score(q,d,u) as Lucene restricts doing so, you have to specify score(qu,d). • Limit the data to high-order queries or divide fields depending on a user as the number of q-u combinations can be enormous. Lucene doc d1 q1u1, q2u2 Lucene doc d2 q2u1, q1u2 53
- 54. Join and Code with Us! Contact us at koji at apache dot org for the details. 54
- 55. Demo or Q & A Thank you! 55