POS
In linguistics, a part of speech (abbreviated POS) is a category into which words are classified based on their grammatical function, syntactic behavior, and semantic role within a sentence.[1] These categories group words that share similar morphological, distributional, and inflectional properties, enabling systematic analysis of language structure.[2] English traditionally recognizes eight primary parts of speech: nouns, pronouns, verbs, adjectives, adverbs, prepositions, conjunctions, and interjections, though some frameworks include additional subclasses such as determiners or articles.[3] Nouns and verbs form the core open classes capable of accepting new members through word formation, while closed classes like prepositions and conjunctions are more stable in inventory.[4] This classification facilitates sentence construction, as words from specific POS combine predictably—for instance, adjectives typically modify nouns, and adverbs qualify verbs.[5] Not all languages align with this model; for example, some lack a distinct adjective class, distributing its functions across verbs or nouns.[4] The concept traces to ancient Greek grammarians like Dionysius Thrax in the 2nd century BCE, who delineated basic categories such as nouns (signifying entities) and verbs (indicating actions or states), with later expansions in Latin and medieval scholarship incorporating adverbs and other forms.[6] Modern linguistics debates the universality of POS systems, with structuralist approaches emphasizing distributional tests (e.g., substitutability in frames) over purely morphological criteria, and computational applications like POS tagging in natural language processing relying on probabilistic models trained on corpora for accuracy rates exceeding 95% in controlled domains.[7] These frameworks underscore POS as a tool for causal analysis of syntax, revealing how word categories drive phrase structure and predicate-argument relations, though cross-linguistic variation challenges rigid universals.[8]Grammar and linguistics
Part of speech
In grammar and linguistics, a part of speech refers to a lexical category into which words are classified based on their syntactic behavior, morphological properties, and semantic roles within sentences.[9] These categories determine how words function in relation to others, such as inflecting for tense or agreeing in number.[10] Classification relies on criteria like distribution in phrases, ability to take affixes, and combinatorial restrictions, rather than isolated meanings.[9] The concept originated in ancient Greek linguistics around the 2nd century BCE, with Dionysius Thrax identifying eight primary categories: noun, verb, participle, article, pronoun, preposition, adverb, and conjunction.[6] This framework influenced Latin grammarians like Priscian, who adapted it for Latin's inflectional system, emphasizing case, gender, and tense markings.[11] By the medieval period, European scholars refined these into systems suited to vernacular languages, though debates persisted over criteria—whether morphological, syntactic, or logical—leading to variations like including interjections or numerals as distinct classes.[12] In English, traditional grammars recognize eight main parts of speech: nouns (naming entities), pronouns (substituting for nouns), verbs (expressing actions or states), adjectives (modifying nouns), adverbs (modifying verbs, adjectives, or other adverbs), prepositions (indicating relations), conjunctions (connecting words or clauses), and interjections (expressing emotion).[10] Some analyses expand this to nine by treating articles as determiners separate from adjectives, reflecting their unique distributional patterns, such as obligatory positioning before nouns in noun phrases.[13] Words can shift categories contextually, as in "run" functioning as noun ("a run") or verb ("to run"), highlighting that part-of-speech assignment depends on sentence-level syntax rather than fixed lexical identity.[14] Cross-linguistically, parts of speech vary significantly; while nouns and verbs appear universal for denoting entities and predications, categories like adjectives are absent in some languages, such as Chinese, where descriptive functions overlap with verbs or nouns.[15] Languages like Japanese distinguish classifiers as a dedicated class for quantifying nouns, unlike English's reliance on numerals or quantifiers.[16] These differences arise from typological features, such as agglutinative morphology in Turkish enabling extensive verb derivations that blur adverbial roles. Empirical studies, including typological databases, confirm that no single inventory applies universally, with analytic languages like Vietnamese minimizing inflectional distinctions across categories.[17]Computing and information technology
Point of sale
A point of sale (POS) system consists of integrated hardware and software that enables merchants to process customer payments, record transactions, and manage operational data such as inventory levels and sales reports.[18][19] These systems serve as the primary interface for retail, hospitality, and service-based transactions, replacing traditional cash registers with digital terminals that handle electronic payments including credit cards, debit cards, and contactless methods.[20][21] The origins of POS systems trace back to mechanical cash registers invented in 1879 by James Ritty to prevent employee theft in saloons, evolving into electronic cash registers (ECRs) by the mid-20th century.[22][23] Computer-based POS systems emerged in 1973 with early terminals connected to mainframes, while IBM contributed in the 1960s by developing standardized coding for supermarket items to streamline checkout processes.[24][25] Touchscreen interfaces appeared in 1985, and cloud-based deployments became widespread after the 2000s, enabling remote access and scalability.[22][26] Core hardware components of a POS system typically include a central processing unit (CPU), monitor for displaying transactions, input devices such as a keyboard, mouse, or touchscreen, and peripherals like barcode scanners for item identification, EMV-compliant card readers for secure chip-and-PIN processing, receipt printers, and customer-facing displays.[27][28] Software layers provide transaction processing, integration with payment gateways, real-time inventory tracking, and reporting capabilities, often syncing with external systems for accounting or customer relationship management (CRM).[29][30] In functionality, POS systems calculate totals, apply taxes and discounts, authorize payments, and generate receipts, while advanced features support loyalty programs, employee management, and omnichannel integration for online-offline sales synchronization.[31][21] By 2025, adoption of cloud-native architectures allows for untethered, mobile POS deployments on tablets or smartphones, reducing reliance on fixed terminals and enabling pop-up or field sales.[32][33] Contemporary POS technologies emphasize artificial intelligence (AI) for predictive analytics, personalized recommendations, and fraud detection, alongside enhanced data integration for business intelligence.[32][34] Systems increasingly incorporate biometric verification, NFC for contactless payments, and API connectivity to third-party services, with market trends favoring scalable solutions for small-to-medium businesses (SMBs) that centralize operations like payroll and inventory.[33][35] Security in POS systems mandates compliance with Payment Card Industry Data Security Standard (PCI DSS), which requires encryption of card data, secure networks, and regular audits to prevent breaches.[36][37] Additional regulations include state-specific rules, such as California's requirements for automatic checkout system inspections and fees, and Americans with Disabilities Act (ADA) guidelines for accessible interfaces with eye-level displays between 43 and 51 inches for wheelchair users.[38][39] Non-compliance risks fines and data exposure, underscoring the need for vendors to prioritize certified, updatable software.[40][41]Part-of-speech tagging
Part-of-speech (POS) tagging is the process of assigning grammatical categories, such as noun, verb, adjective, or adverb, to each word in a text corpus, based on the word's definition, its context within the sentence, and surrounding words.[4] This task reveals syntactic structure and disambiguates words with multiple possible roles, like "fish" as a noun or verb.[4] POS tagging serves as a foundational step in natural language processing (NLP), enabling higher-level analyses by providing lexical and contextual cues about word relationships.[42] Early POS tagging relied on rule-based systems, where human-crafted linguistic rules determined tags based on morphological features and fixed patterns.[4] Statistical methods emerged in the 1990s, with hidden Markov models (HMMs) becoming dominant; these probabilistic models estimate tag sequences by modeling transitions between tags and emissions of words given tags, achieving 96.7% accuracy on the Penn Treebank Wall Street Journal (WSJ) corpus using trigrams as reported by Brants in 2000.[4] Conditional random fields (CRFs), introduced around 2001, improved upon HMMs by directly modeling conditional probabilities of tag sequences given observations, handling dependencies more effectively without independence assumptions inherent in generative HMMs.[43] Neural network approaches, particularly bidirectional long short-term memory (BiLSTM) networks combined with CRFs, have since pushed accuracies higher, often exceeding 97% on standard English benchmarks like the Penn Treebank's WSJ sections (typically sections 0-18 for training and 19-21 for testing with 45 tags).[44] The Penn Treebank, developed in the early 1990s from WSJ articles, remains the primary benchmark for English POS tagging evaluation, with performance measured by per-word accuracy.[4] State-of-the-art systems, including deep learning models like convolutional neural networks or transformers fine-tuned on this corpus, routinely achieve 97-98% accuracy, though error rates persist for ambiguous or rare words.[44] Challenges include handling out-of-vocabulary words, context-dependent ambiguities (e.g., prepositions versus particles), and language-specific morphological complexity, which rule-based and early statistical methods addressed poorly but neural models mitigate through contextual embeddings.[4] POS tagging underpins numerous NLP applications, including syntactic parsing to build phrase structures, machine translation by aligning grammatical roles across languages, and information extraction for identifying entities and relations.[4] It aids sentiment analysis by distinguishing opinion-bearing adjectives from others, enhances search engines through query understanding, and supports text simplification by restructuring sentences based on tagged components.[45] In low-resource languages, hybrid neural-CRF models adapt tagged data to improve downstream tasks like named entity recognition.[46] Despite advances, tagging accuracy directly impacts these systems' reliability, with empirical studies showing cascading errors in parsing reduced by 20-30% via improved taggers.[4]Navigation
Point of sailing
In nautical navigation, a point of sailing denotes the directional relationship between a sailing vessel's heading and the true wind direction over the water surface, which determines sail trim, boat speed, and maneuverability.[47] This concept is fundamental to sailboat handling, as vessels cannot sail directly into the wind (known as being "in irons" at 0 degrees), requiring tacking maneuvers to make progress upwind.[48] Points of sailing are categorized broadly into upwind (close-hauled), reaching (intermediate angles), and downwind (running), with optimal performance varying by vessel type, wind strength, and sea state.[49] The primary points of sailing, defined by approximate angles relative to the true wind, are as follows:| Point of Sailing | Approximate True Wind Angle | Key Characteristics |
|---|---|---|
| In Irons | 0° | Vessel stalled directly head-to-wind; sails luff and provide no forward drive, necessitating backing or maneuvering to escape.[50] |
| Close-Hauled | 30°–45° | Sails trimmed flat and tight; maximum upwind progress with heel angle; typical speeds limited by hull design and wind velocity.[51][52] |
| Close Reach | 45°–60° | Sails eased slightly from close-hauled; faster than upwind but requires vigilant wind shifts to avoid accidental jibing.[51] |
| Beam Reach | 90° | Wind abeam; sails at roughly halfway outhaul; often the fastest point due to balanced power and minimal drag.[47][50] |
| Broad Reach | 120°–135° | Wind aft of beam; sails further out; high speeds possible but increased risk of broaching or accidental gybe in gusts.[52][51] |
| Running | 135°–180° | Downwind with wind from astern; sails winged out or using spinnaker; slowest relative speed as apparent wind decreases, prone to rolling.[48][49] |
Arts and entertainment
P.O.S. (musician)
Stefon Leron Alexander (born August 18, 1981), known professionally as P.O.S., is an American hip hop artist based in Minneapolis, Minnesota.[53][54] A co-founder of the independent hip hop collective and record label Doomtree, he has released music through Rhymesayers Entertainment since the mid-2000s, blending rap with punk rock elements in an experimental style.[55][56] His work draws from his early involvement in punk bands such as Building Better Bombs and Cadillac Blindside, transitioning into hip hop via collaborative projects like Cenospecies.[57][54] P.O.S. debuted with the self-released album Ipecac Neat in 2004, followed by Audition in 2006 under Rhymesayers, establishing his reputation for dense, lyrical content infused with rock influences.[58] His 2009 release Never Better marked a commercial breakthrough, earning praise for its energetic production and personal themes, including tracks produced by collaborators like Ant.[59][60] Subsequent albums include We Don't Even Live Here (2012) and Chill, dummy (2017), the latter reflecting on health struggles following a kidney transplant in 2014 that nearly ended his career.[55] He has also contributed to side projects like Four Fists with Astronautalis and appeared on compilations tied to the Minneapolis indie rap scene.[54][53] In June 2020, amid broader allegations of misconduct in the Twin Cities music community, P.O.S. faced accusations from multiple women of emotional abuse, including gaslighting, lying, and mistreatment in relationships; he issued a public apology acknowledging these claims and announced a temporary step away from music.[61][62][63] The statement, prompted by posts from Doomtree affiliate Dessa, emphasized his intent to address the harm caused without denying the reports.[61] He resumed activity with the EP RELAY in 2024, signaling a return after personal reflection and recovery.[64]Discography
- Ipecac Neat (2004)[57]
- Audition (2006)[58]
- Never Better (2009)[59]
- We Don't Even Live Here (2012)[54]
- Chill, dummy (2017)[55]
- RELAY (EP, 2024)[64]