AUTOMATED DISCOVERY OF TELIC RELATIONS FORWORDNETMARCO DE BONISURESH MANANDHARBackground TermsGlosses – Every word and synset in WordNet contains ashort (about one sentence long) descriptioncalled glosses.Telic Relations - The goal / function of an object. Thepurpose an agent has in performing an act or thebuilt-in function.IntroductionBackground Terms (cont)Telic Relations (cont) – Some examples of telic relations,• The telic of “milk” might be “drink”.• The telic of “wood” might be either“burn” (make a fire) or build (make furniture). - Objects may have none, just one, or many telic relationships.IntroductionGoal of Article Telic relations are talked about inWordNet (5 papers, pg. 18, 22 ),but never actually implemented. The Goal of the article is to presentan algorithm to automaticallydiscover telic relations of words andsynsets by looking at glosses (thedescriptions of words).IntroductionGeneral algorithmFor every noun in WordNet1. Parse out a telic relation(s) from theglosses for the noun.2. Since the telic word(s) will usuallyhave many senses, find theappropriate synset to match up thenoun to.IntroductionThe Algorithm (Pt I: finding the telic word)The Algorithm (Pt I: finding the telic word)- By looking at special patterns (specific- By looking at special patterns (specific wording) within the glosses, telic words wording) within the glosses, telic words can be parsed out. can be parsed out.- The patterns used were- The patterns used wereTo deceive someoneTo deceive someone““the use of tricks to deceive someone.the use of tricks to deceive someone.””TrickeryTrickery“…“… use of use of …… to toTelic.Telic.””Stimulation of theStimulation of theHeartHeart““Obtained from a number of plants andObtained from a number of plants andused to stimulate the heart used to stimulate the heart …”…”CardiacCardiacglycosideglycoside“…“… used to Telic. used to Telic.””Trace drawingsTrace drawings““a semitransparent paper used fora semitransparent paper used fortracing drawings.tracing drawings.””Tracing paperTracing paper“…“… used for Telic. used for Telic.””Detect breast tumorsDetect breast tumors““a diagnostic procedure to detect breasta diagnostic procedure to detect breasttumors tumors by use ofby use of X rays. X rays.””MammographyMammography“…“… to TELIC_VERB to TELIC_VERBby use of by use of …”…”telictelicGlossGlossEx. WordEx. WordPatternPatternUse as a statelyUse as a statelyornamental.ornamental.““Large evergreen Large evergreen …… used in wet warm used in wet warmregions as a stately ornamental.regions as a stately ornamental.””Giant taroGiant taro“…“… used in used in …… as a as aTELIC.TELIC.””Dwelling.Dwelling.““a barge that is designed and equipped fora barge that is designed and equipped foruse as a dwelling.use as a dwelling.””HouseboatHouseboat“…“… for use as for use asTELIC.TELIC.””Racing,Racing,transportation oftransportation ofpassengers.passengers.““light rowboat for use in racing or forlight rowboat for use in racing or fortransporting goods and passengers intransporting goods and passengers ininland waters and harbors.inland waters and harbors.””WherryWherry“…“… for use in for use in ……TELIC_ING-VERB.TELIC_ING-VERB.””Making soap,Making soap,dressing leather anddressing leather andlubrificationlubrification..“…“… from seal blubber; used in making soap from seal blubber; used in making soapand dressing leather and as a lubricant.and dressing leather and as a lubricant.””Seal oilSeal oil“…“… used in used inTELIC_ING-VERB.TELIC_ING-VERB.””ModelingModeling“…“… resembling clay; used as a substitute resembling clay; used as a substitutefor clay or wax in modeling.for clay or wax in modeling.””PlasticinePlasticine“…“… used as used as …… in inTELIC_ING-VERB.TELIC_ING-VERB.””telictelicGlossGlossEx. WordEx. WordPatternPattern- The patterns used were- The patterns used wereThe Algorithm (Pt I: finding the telic word)The Algorithm (Pt I: finding the telic word)Problems.Problems. - Sometimes its hard to pick one specific - Sometimes its hard to pick one specificnoun phrase or verb phrase for the telicnoun phrase or verb phrase for the telicword.word. - Over-generalization: When words like - Over-generalization: When words like““bebe””, , ““dodo””, , ““makemake””, , ““thingthing”” were the telic were the telicrelations, more specific words had to berelations, more specific words had to befound.found.The Algorithm (Pt I: finding the telic word)The Algorithm (Pt I: finding the telic word)The telic word(s) of an object will most likely belongto more than one synset. The second part of thealgorithm is to determine the appropriate synset forthe object by looking at the distance (or difference)between different concepts. There are many different approaches that can beuse to find distance between concepts ( forexample wnconnect). The authors measured therelationship between two concepts with “Sematicdistance.”The Algorithm (pt II: finding the correct synset)Given a object w and its telic t, the semanticdistance takes all glosses from each sense of tand compares it to the gloss of w (the object).The semantic distance is a sigmoid function thatapplies an number to each sense of t (the telic). To calculate the actual sematic distance theauthors look at all the words surrounding theglosses in t (the telic), and words in gloss w (theobject).The Algorithm (pt II: finding the correct synset)Semantic DistanceGiven object w with gloss GWw, it has a telic t withT representing all possible synsets (senses) of t.We want to find the correct synset, ts T, by: ts = Argmaxts’ T sd( GWw, ts’)* Where sd( a, s ) calculates the semantic distancebetween sentence a and word w.The Algorithm (pt II: finding the correct synset)Finding the Correct Synset ts*I have no clue what exactly Argmax means∈∈In Order to find semantic distance sd for each ts’T, we take all the words in the gloss of ts’ (call itset GT), plus all the words in the glosses of thehypernyms and hyponyms of ts’ to the depth of3, creating the set TSts. So TSts = { w | w GT V w hyperg(ts’, 3) V w hypog(ts’, 3) }.The Algorithm (pt II: finding the correct