book

Practical Natural Language Processing

by Sowmya Vajjala, Bodhisattwa Majumder, Anuj Gupta, Harshit Surana

June 2020

Beginner to intermediate

456 pages

13h 6m

English

O'Reilly Media, Inc.

Book available

Read now

Unlock full access

Includes

Has Sandbox

Why We Wrote This BookThe PhilosophyScopeWho Should Read This BookWhat You Will LearnStructure of the BookHow to Read This BookConventions Used in This BookUsing Code ExamplesO’Reilly Online LearningHow to Contact UsFurther InformationAcknowledgments
NLP in the Real WorldNLP TasksWhat Is Language?Building Blocks of LanguageWhy Is NLP Challenging?Machine Learning, Deep Learning, and NLP: An OverviewApproaches to NLPHeuristics-Based NLPMachine Learning for NLPDeep Learning for NLPWhy Deep Learning Is Not Yet the Silver Bullet for NLPAn NLP Walkthrough: Conversational AgentsWrapping Up
Data AcquisitionText Extraction and CleanupHTML Parsing and CleanupUnicode NormalizationSpelling CorrectionSystem-Specific Error CorrectionPre-ProcessingPreliminariesFrequent StepsOther Pre-Processing StepsAdvanced ProcessingFeature EngineeringClassical NLP/ML PipelineDL PipelineModelingStart with Simple HeuristicsBuilding Your ModelBuilding THE ModelEvaluationIntrinsic EvaluationExtrinsic EvaluationPost-Modeling PhasesDeploymentMonitoringModel UpdatingWorking with Other LanguagesCase StudyWrapping Up
Vector Space ModelsBasic Vectorization ApproachesOne-Hot EncodingBag of WordsBag of N-GramsTF-IDFDistributed RepresentationsWord EmbeddingsGoing Beyond WordsDistributed Representations Beyond Words and CharactersUniversal Text RepresentationsVisualizing EmbeddingsHandcrafted Feature RepresentationsWrapping Up
ApplicationsA Pipeline for Building Text Classification SystemsA Simple Classifier Without the Text Classification PipelineUsing Existing Text Classification APIsOne Pipeline, Many ClassifiersNaive Bayes ClassifierLogistic RegressionSupport Vector MachineUsing Neural Embeddings in Text ClassificationWord EmbeddingsSubword Embeddings and fastTextDocument EmbeddingsDeep Learning for Text ClassificationCNNs for Text ClassificationLSTMs for Text ClassificationText Classification with Large, Pre-Trained Language ModelsInterpreting Text Classification ModelsExplaining Classifier Predictions with LimeLearning with No or Less Data and Adapting to New DomainsNo Training DataLess Training Data: Active Learning and Domain AdaptationCase Study: Corporate TicketingPractical AdviceWrapping Up
IE ApplicationsIE TasksThe General Pipeline for IEKeyphrase ExtractionImplementing KPEPractical AdviceNamed Entity RecognitionBuilding an NER SystemNER Using an Existing LibraryNER Using Active LearningPractical AdviceNamed Entity Disambiguation and LinkingNEL Using Azure APIRelationship ExtractionApproaches to RERE with the Watson APIOther Advanced IE TasksTemporal Information ExtractionEvent ExtractionTemplate FillingCase StudyWrapping Up
ApplicationsA Simple FAQ BotA Taxonomy of ChatbotsGoal-Oriented DialogChitchatsA Pipeline for Building Dialog SystemsDialog Systems in DetailPizzaStop ChatbotDeep Dive into Components of a Dialog SystemDialog Act ClassificationIdentifying SlotsResponse GenerationDialog Examples with Code WalkthroughOther Dialog PipelinesEnd-to-End ApproachDeep Reinforcement Learning for Dialogue GenerationHuman-in-the-LoopRasa NLUA Case Study: Recipe RecommendationsUtilizing Existing FrameworksOpen-Ended Generative ChatbotsWrapping Up

Search and Information RetrievalComponents of a Search EngineA Typical Enterprise Search PipelineSetting Up a Search Engine: An ExampleA Case Study: Book Store SearchTopic ModelingTraining a Topic Model: An ExampleWhat’s Next?Text SummarizationSummarization Use CasesSetting Up a Summarizer: An ExamplePractical AdviceRecommender Systems for Textual DataCreating a Book Recommender System: An ExamplePractical AdviceMachine TranslationUsing a Machine Translation API: An ExamplePractical AdviceQuestion-Answering SystemsDeveloping a Custom Question-Answering SystemLooking for Deeper AnswersWrapping Up
ApplicationsUnique ChallengesNLP for Social DataWord CloudTokenizer for SMTDTrending TopicsUnderstanding Twitter SentimentPre-Processing SMTDText Representation for SMTDCustomer Support on Social ChannelsMemes and Fake NewsIdentifying MemesFake NewsWrapping Up
E-Commerce CatalogReview AnalysisProduct SearchProduct RecommendationsSearch in E-CommerceBuilding an E-Commerce CatalogAttribute ExtractionProduct Categorization and TaxonomyProduct EnrichmentProduct Deduplication and MatchingReview AnalysisSentiment AnalysisAspect-Level Sentiment AnalysisConnecting Overall Ratings to AspectsUnderstanding AspectsRecommendations for E-CommerceA Case Study: Substitutes and ComplementsWrapping Up
HealthcareHealth and Medical RecordsPatient Prioritization and BillingPharmacovigilanceClinical Decision Support SystemsHealth AssistantsElectronic Health RecordsMental Healthcare MonitoringMedical Information Extraction and AnalysisFinance and LawNLP Applications in FinanceNLP and the Legal LandscapeWrapping Up
Revisiting the NLP Pipeline: Deploying NLP SoftwareAn Example ScenarioBuilding and Maintaining a Mature SystemFinding Better FeaturesIterating Existing ModelsCode and Model ReproducibilityTroubleshooting and InterpretabilityMonitoringMinimizing Technical DebtAutomating Machine LearningThe Data Science ProcessThe KDD ProcessMicrosoft Team Data Science ProcessMaking AI Succeed at Your OrganizationTeamRight Problem and Right ExpectationsData and TimingA Good ProcessOther AspectsPeeking over the HorizonFinal Words