Natural language processing deals with understanding text and spoken words as a human would. It is a fundamental component of many human/machine interactions (vocal assistants, dictation software, voice-operated systems, ...), text processing, analysis (suggestions, keyword spotting, translation, ...) and much more. The Wolfram Language natural language processing functionality is a combination of rule-based and machine learning language models, including LLMs. It builds on top of advanced text mining and string manipulation capabilities and is integrated with a large visualization suite and extensive built-in linguistic data.

Text Generation & Acquisition

LLMSynthesize — generate text from a prompt using an LLM

TextRecognize  ▪  ResourceData  ▪  WikipediaData

Import — import text from files or the web

"Text"  ▪  "PDF"  ▪  "HTML"  ▪  "CSV"  ▪  ...

LLM-Based Operations »

LLMResourceFunction — apply operations from the Wolfram Prompt Repository

LLMFunction — apply operations specified by natural language descriptions

LLMPromptGenerator — add context-dependent messages to an LLM prompt

LLMPrompt  ▪  LLMTool  ▪  ChatEvaluate  ▪  ...

Text Mining

SemanticSearch — search based on the contextual meaning of terms

TextSearch — search an index or directory, returning a list of documents

Find, FindList — search files for records containing particular strings

StringTake  ▪  StringReplace  ▪  StringCases  ▪  RegularExpression  ▪  ...

Text Normalization »

RemoveDiacritics — remove diacritics such as accents, umlauts, etc.

CharacterNormalize — reduce or decompose characters to normal forms (e.g. ¼ to 1⁄4)

TextTranslation  ▪  Transliterate  ▪  DeleteStopwords  ▪  WordStem  ▪  ToLowerCase  ▪  ...

Tokenization

StringSplit — split a string at spaces or other delimiters

StringCases — find cases of string patterns

TextCases  ▪  TextSentences  ▪  TextWords  ▪  TextStructure

Feature Extraction

FeatureExtraction — extract numerical features from text

NetModel — pre-trained networks for text feature extraction

"GloVe"  ▪  "BERT"  ▪  "ELMo"  ▪  "GPT2"  ▪  ...

NetGraph  ▪  LongShortTermMemoryLayer  ▪  AttentionLayer

"Tokens"  ▪  "SubwordTokens"  ▪  "Characters"  ▪  ...

Content Extraction

TextSummarize — automatically produce different types of summarization

FindTextualAnswer — attempt to find answers to questions from text

TextContents, TextCases, TextPosition — extract semantic elements in text

Text Classification

Classify — classify strings based on training data or built-in classifiers

"Language"  ▪  "Profanity"  ▪  "Sentiment"  ▪  ...

LanguageIdentify — identify what language a text is in

Text Clustering

FindClusters — find clusters in string data

ClusteringTree  ▪  ClusteringComponents  ▪  ClusterClassify

Text Analysis »

WordCounts — count of words or -grams

CharacterCounts  ▪  WordFrequency  ▪  WordData  ▪  PartOfSpeech  ▪  ...

Text Visualization

WordCloud — generate a word cloud from word frequencies or weights

Snippet — extract a snippet of text

Style, Highlighted — style text with color, font, size, background, etc.