Class Note for CMPSCI 585 at UMass(9)
Class Note for CMPSCI 585 at UMass(9)
Popular in Course
Popular in Department
This 49 page Class Notes was uploaded by an elite notetaker on Friday February 6, 2015. The Class Notes belongs to a course at University of Massachusetts taught by a professor in Fall. Since its upload, it has received 18 views.
Reviews for Class Note for CMPSCI 585 at UMass(9)
Report this Material
What is Karma?
Karma is the currency of StudySoup.
You can buy or earn more Karma at anytime and redeem it for class notes, study guides, flashcards, and more!
Date Created: 02/06/15
Course Overview Lecture 1 Computational Linguistics CMPSCI 585 Fall 2007 University of Massachusetts Amherst Andrew McCaIIum httpwwwcsumassedumccallumcoursesinlp2007 Where to find syllabus announcements slides homeworks Andrew McCallurn UNIass Amherst including material from Chris Manning and Jason Eisner Today s Main Points Why is natural language interesting and difficult complex and ambiguous Why How to we resolve this ambiguity Six layers of natural language Natural Language Processing overview current successes Get to know each other and our motivations for being here Course mechanics what you can expect Andrew McCallum UMass Amherst including material from Chris Manning and Jason Eisner 1967 Stanley Kubrick Arthur C Clarke filmmaker author futurist 1928 1999 1917 Andrew McCallum UMass Amherst ncludjng material from Chris Manning and Jason Eisner Andrew McCallum UMass Amherst including material from Chris Manning and Jason Eisner HAL s Capabilities Display graphics Play chess Natural language production and understanding Vision Planning Learning mmmmmmmmmmmmmmmmmmmm st Q mm MONTY NEWBDRN KASPARO I Andrew McCallum UMass Amherst including material from Chris Manning and Jason Eisner Natural Language Understanding Now David Bowman Open the pod bay doors Hal HAL I m sorry Dave I m afraid I can t do that David Bowman What are you talking about Hal HAL I know that you and Frank were planning I to disconnect me and I39m afraid that39s something I cannot allow to happen Many useful tools but none that come even close to HAL s ability to communicate in natural language Andrew McCallurn UNIass Amherst including material from Chris Manning and Jason Eisner 1950 Alan Turing 1912 1954 Turing Test Computing Machinery and Intelligence Mind Vol 59 No 236 pp 433460 1950 I propose to consider the question quotCan machines think We can only see a short distance ahead but we can see plenty there that needs to be done Andrew McCallum UMass Amherst including material from Chris Manning and Jason Eisner Layers of Computational Linguistics Phonetics amp Phonology Morphology Syntax Semantics Pragmatics Discourse FDP J F N mmmmmmmmmmmmmmmmmmmm st 1 Phonetics amp Phonology The study of language sounds systems of discrete how they are sounds eg languages physically formed syllable structure It is easy to recognize speech JeetJet It is easy to wreck a nice beach Andrew McCallum UMass Amherst 39 ding material from Chris Manning and Jason Ersner 2 Morphology The study of the subword units of meaning H not to attach Even more necessary in some other languages eg Turkish uygarasz iramadikarimizdanmissinizcasina uygar las tir ama dik lar imiz dan mis sinz casina behaving as if you are among those whom we could not civilize Andrew McCanum UMass Amherst including material from Chris Manning and Jason Eisner 3 Syntax The study of the structural relationships between words N V SBAR SBAR S I know that NP VP I NP CONJ NNP i r J you and Frank were planning to disconnect me Not same structure Andrew McCanum UMass Amherst including material from Chris Manning and Jason Eisner 4 Semantics The study of the literal meaning ACTION disconnect ACTOR you and Frank OBJECT me Andrew McCanum UMass Amherst 39 uding material from Chris Manning and Jason Eisner 5 Pragmatics The study of how language is used to accomplish goals What should you conclude from the fact I said something How should you react I in Includes notions of polite and indirect styles Andrew McCallum UMass Amherst including material from Chris Manning and Jason Eisner 6 Discourse f Linn than The structure of conversations turn taking thread of meaning Open the pod bay doors Hal I m sorry Dave I m afraid I can t do that What are you talking about Hal I know that you and Frank were planning to disconnect me and I39m afraid that39s something I cannot allow to happen Andrew McCallum UMass Amherst including material from Chris Manning and Jason Eisner Linguistic Rules Eg Morphology To make a word plural add s mmmmmmmmmmmmmmmmmm st It 5 dog 9 dogs baby 9 babies dish a dishes goose a geese child 9 children fish a fish l Inherent Ambiguity in Syntax 1 1 NY Times headline 17 May 2000 S NP VP I N NNP V NP NP PP Fed raises NN NN CD NN PP NP interest rates 05 in NN VP effort V VP to V NP control N N Andrew McCallum UMass Amherst including material from Chris Manning and Jason Eisner Where are the ambiguities Partofspeech ambiguities SyntaCtiC attaChment ambiguities VB VBZ VBZ VBZ NNP NNS NNS NNS CD NN Word sense ambiguities Fed 4 federal agent interest ea feeling of wanting to know or learn more Semantic interpretation ambiguities above the word level Andrew McCallum UMass Amherst including material from Chris Manning and Jason Eisner Effects of VN Ambiguity 1 S A NP VP I NNP V NP l Fed raises NN NN interest rates Andrew McCanum UMass Amherst 39 ding material from Chris Manning and Jason Eisner Effects of VN Ambiguity 2 s NP VP N N v NP ii I Fed raises interest N rates Andrew McCanum UMass Amherst 39 ding material from Chris Manning and Jason Eisner Effects of VN Ambiguity 3 S NP VP N N N v NP I i l Fed raises interest rates CD N 05 Andrew McCanum UMass Amherst 39 ding material from Chris Manning and Jason Eisner Ambiguous Headlines Iraqi Head Seeks Arms Juvenile Court to Try Shooting Defendant Teacher Strikes Idle Kids Stolen Painting Found by Tree Kids Make Nutritious Snacks Local HS Dropouts Cut in Half British Left Waffles on Falkland Islands Red Tape Holds Up New Bridges Clinton Wins on Budget but More Lies Ahead Ban on Nude Dancing on Governor s Desk mmmmmmmmmmmmmmmmmmmm st Language Evolves Morphology We learn new words all the time bioterrorism cyberstalker infotainment thumb candy energy bar Partofspeech Historically kind and sort were always nouns knowe that sorte of men ryght well 1560 Now also used as degree modi ers I m sort of hungry Present It sort 0 stirs one up to hear about old times 1833 mmmmmmmmmmmmmmmmmmmm st Natural Language Computing is hard because Natural language is highly ambiguous at all levels complex and subtle fuzzy probabilistic interpretation involves combining evidence involves reasoning about the world embedded a social system of people interacting persuading insulting and amusing them changing overtime mmmmmmmmmmmmmmmmmmmm st Probabilistic Models of Language To handle this ambiguity and to integrate evidence from multiple levels we turn to The tools of probability Bayesian Classifiers not rules Hidden Markov Models not DFAs Probabilistic Context Free Grammars other tools of Machine Learning Al Statistics mmmmmmmmmmmmmmmmmmmm st Another Area where Probabilistic Combination of Evidence Won Andrew McCanum UMass Amherst including material from Chris Manning and Jason Eisner Natural Language Processing Natural Language Processing NLP studies 5 9 how to get computers to do useful things with 63 natural languages 6amp8 Most commonly Natural Language Understanding The complementary task is Natural Language Generation NLP draws on research in Linguistics Theoretical Computer Science Artificial Intelligence Mathematics and Statistics Psychology Cognitive Science etc mmmmmmmmmmmmmmmmmmmm st What amp Where is NLP Goals can be very farreaching True text understanding Reasoning and decisionmaking from text Realtime spoken dialog Or very downtoearth Searching the Web Contextsensitive spelling correction Analyzing readinglevel or authorship statistically Extracting company names and locations from news articles These days the later predominate as NLP becomes increasingly practical focused on performing measurably useful tasks now Although language is complex and ambiguity is pervasive NLP can also be surprisingly easy sometimes rough text features often do half the job Andrew McCallum UMass Amherst including material from Chris Manning and Jason Eisner Linguistics Linguistics is the study of natural languages Understanding this naturallyoccurring 9 phenomenon 000ch Structure meaning how acquired differences and commonalities across languages Linguistics draws on research in Natural Language Processing Theoretical Computer Science Artificial Intelligence Mathematics and Statistics Psychology Cognitive Science etc mmmmmmmmmmmmmmmmmmmm st Andrew McCallun including material Example Applications of NLP C Google Search natural language processing 1 7 v at a i i 4 r L5 httpfwwwgooglexomlsearcmhl en xIe ISO BSSQ l tec QvLucclc Yahoo Google Slashdot Newsv McC Researchv Reviewingv Macv Java Thesaurus Fundingv Teachingv 7 l0 Jungle Search natural I Language Tools Search Tigs l Google Search Advanced Search Preferences 300816quot natural language processing Searchedthe web for natural Ian u e rocessin Results 1 10 of about 2210000 Search took 021 seconds Natural Lan ua e Processin Natural Language Processing should make it possible for people to use computers in much the same way that they would use a human assistant to get their work researchmicrosoftcomlnlpf 28k Cached Similar pages Spol isored Links Natural Language Search Returns n orc rolerant searches Installs in d s Free White Papers wwwprimuscom Interest ISI s Natural Language Groug Overview of Research Environment Natural Language Processing at USCJISI USC offers a wide range of courses in areas related to natural language processing Description The Natural Language Processing group at the Information Sciences Institute of the University of Southern Categch Computers Artificial Intelligence gt gt Reseanzli Groups wwwisiedulnaturaltlanguagefnlpat islhtml a 15k Cached Similar gates lNLP News All the news that s fit to parse Hun39an Language Technwugf fleldmethods net Interest Foundations of Statistical Natural Language Processing Foundations of Statistical Natural Language Processing Chris Manning and Hinrich SchLitze Foundations of Statistical Natural Language Processing MIT Press nlpstanfordedulfsnlpt 7k Cached Similar gages Natural Lang Processing Text Mlquotng sea Ln NLPV For SCIEI IllfID rc analyse mariadnegenomicscom Interest Yahool Directorv Arti cial Intelligence gt Natural Language Artificial Intelligence gt Natural Language Processing Directory gt Science gt Computer Science gt Artificial Intelligence gt Natural Language Processing rtir Iahnn rnmthinnrnlr nmmitnr SrinnrnlArtifir inl Intnllinnnrni39 Natural Innmmnn Prnr ncainnl l Work at Gale 1 Iciv39 ririn EKDEI IC39TI Example Applications of NLP MSWord spelling correction grammar checking Ifjmu39use 39I39LI39IinEIS n 39Wurdjrpuhavr nn duutrbnptic ed re 21 anymis 5 Elle dwurds 101 tDheeactal 39WUI dS39th didyeulamwthatynucancarts cttheseermrssimply I39u IiEI IZISD WDI d E i 39gie39FEIU39 39liSt Df thE39WDde39th t39i wordgrouwmtWmthe listjgmusimplypickitf39 i appease apparels appeals appear Lgnure Fill add AutuCorrect Ir 3 Spelling Andrew McCallum UMass Amherst including material from Chris Manning and Jason Eisner mmcam Example Applications of NLP 608 Coogle News r Google News New Searcn Hrsrory39nou r KWMewsgaog mePaumDUAMG6MMIvlZPZZe3kAneETszXlkNPSKNDuumV KJWBr v mlnxsuDPnn Aszlmd AsnsReu WWWRmmws muons Auzonsrlugm ooogue yanoar Run Web Images Groups News Frucg e Luca mare Advanud Searzh News Searzh the Web rimudessoogre Nous L m movel sranoaro News 1 Yex Versmn TopSloriesus j Cu VMematha r reraro Tnbune Mos Popular all 1 512 related E News Moms uote Rememberin Core a Scot Kin seame Fast Intelligence 72 hnurs ago By THE ASSOCIATED PRESS Comments by inends and RSS Atom Aoom Feeds World I U3 Nita Seen as Car in the Torch of Rea an Lns Angeles nmes r a hams ago Bus 55 By mayo 6 Savage rrmes Staff Wnrer WASHINGTON 7 sirreeh Twenlyshve years ago Presroem Reagan came to Washmg on I spans wun ooru orans 1 move me Supreme Com m ore ngm Ne ano rus awyers wanted a rugn ooun Emenainmem AW Faces Senare Con rmalmn Vol FOX News I Hem Senate re eors Democraks move to mock Ahm s oonnnnauon Washmgmn Push Gas News VABC News 7 Guaroran umrmu Amovgenerared 1o minures agn Personalize this page Rates In se as Feo balances an BBCNews H 1Hvealed Maoanna Meeqs sage MSNECV 5H 52 mated Sleelers know how to WIN Deles com 7 3 ma rerareo Ed 39 Haggis niek nabs s Oscar nnms Londnn Free Press 7 5 M9 ve ated nh Daleas CNN 5 34912 aled led Glabal Burden areene e e lo vote a a sl ran NDTVum r 37 mmures ago Inora rs ukery to vote agamst Iran at me Intemauonar Atamm Energy Agenoy IAEA meeung m Vrenna on February 2 ms comes after Russra ano cruna wrnon abstamed rom young a me IAEA meeung m SepremberZDDS onangeo merrsrano Iran warns of end to nuc eardm omanvMaH amp GuavmarLthne Dane mnhldlrg mam In39ured ABC newsmen return to US Munsiers and Crlucsvcam 727 mlm es ago LANDSTUHL Germany um 7 ABC newsman Bob Wooorurr ano cameraman Doug Vogr eft a nosouar m Germany or semesoa Mo on Tuesoay a ev bemg unwed by a bomb m Iraq he men were armed ro the Landsmm admvers oi Coreua Soon ng me wroou of me Rey Mamn quot The News Arm Lumer ng Ir rouomng neroeam on Tuesday J n s a break Bub Woodmff on Cava aware News momrng for me ano ror many oeoore Corena Soon ng Emck Inyrrauanar Mourners yrsrr ng s tomb a gay respects 0 ms wmow Access Norm Georgra Wenoy Wassevsrem Skate or me Umzm oryu Rrgrus can Coretta Scott ng 192mm Demncracv Nw Exxon Mob Saddam Hussem WJLA r aosNewerr XIAVI V e Pageanemoom e an 353 relareo Doug Vogr Soreen Actors 6qu World x mm U eon Aec syyooorufr rs recoyannn DeJNews om i Example Applications of NLP Information Extraction Find experts employees wwwzoominfocom http Dr Andrew McCallum Action Editor Journal of Machine Leamlng Research Last Mentioned on 10f1212003 Actions JMLR Inc E Send This Profile Update Your Profile litigllwvvamlr org mall Hal Available other Titles Held Member Editorial Board Additional Current Employment Carnegie Mellon University Post Doctoral Fellow Agunct Faculty Member JMLR which publishes highouality scholarly articles in all areas of machine learning competes with the commercial journal Machine Learning which costs US1DOBt A number of Machine Learning editorial board members have resigned to join the editorial board of JMLR more Adjunct Faculty Position Click here to find other people who University of Massachusetts Amherst c0 Research Associate Professor work for JMLR Inc Adjunct Faculty Research Scientist WhizBanu Labs Inc Board Memberships and Affiliations Intelliseek Inc Member of Advisory Board IJCAI Member Program Committees past AAAI Member Program Committees past ICML Member Program Committees past Contact Us Corporate Headquarters 3210 North Canyon Road Suite 200 Provo UT 84604 Phone 801 413 7100 Fax 601 81H300 ht g rvwwwhlzbangabscom NiPS Member Program Committees past Past Employment History WhizBang Labs Inc Vice President of Research and Development Just Research Research Scientist Biomedical Information Communication Center of Oregon Health Sciences University Machine Learning Researcher WhizBang Labs founded in 1999 is a leader in the field of lnforrnation extraction and document autotagging from unstructured data sources Through our products and services we analyze unstructured content in both online and offvline formats locate and extract key data elements into XML tagged lmorei Education Click here to find other people who work for WhizBang Labs Inc University of Rochester PhDt Computer Science Dartmouth College Bachelor of Arts Computer Science Information about Andrew McCallum was compiled from 6 sources Intelliseek Inc 1128 Main Street Fourth Floor Cincinnati OH 452027236 Phone 513418 6700 Go to ti ttp networking 2 eliyo n com I Networking Idefaultasp39 Andrew McCallum UMass Amherst including material from Chris Manning and Jason Eisner Example Applications of NLP Information Extraction Job Openings ampusungnaunnaun anaxeculwSearchlnmlncusmguniheFiud cm guy manIMHHaEX r V Wf foodsdencuomhhz rm Sir 2p mun INTERNATWI m i 5am in Am manm UMass mm i his in i i Adam you mung View Favnule Haiv Tiile Ice Cream Guru 39rmwm 5m Rehexh Hume Seaich Favnu player foodsciencecorn J obCategory TravelHospitality JobFunciion Food Services obLocation Upper MidWest Contact Phone 8004882611 DaleExtracted January 8 2001 Abam urns maximum Mainrfwd manuf la lL ml l39arm Job nun mm mm um Scum m Help lndmry man my 0mm in sin Map emu Source WWWfoodsciencecomjobsgnidwesthtn foddscienceComJob1 mm mm mchns Mzmlrg ma yamme Example Applications of NLP Information Extraction Job Openings meee a hug me Vhpdag camhams mm F dJoh You Autumn Resume Ceer 39 Summrt 39 Emplnyers mm waeea Shaw2552 Jabs navmllu Nnvm o Heakh Care m w 2 Heakh Care m MD 1 e Sa es m Nquot 3 547514 quot M m V Cameaana m w e Jub uppmmnmes Comm m MD ma a Human rmm 53 641 Emmevers Campanaaaan Wear based HR aavaaamaa F nd a JD M W SW quot5 and Customev Head Foo39baH Cuazh Manaaamam Systems my raa aauaaaah LaamMaya Am FombaH Cuazh We mede wta ma Smoo FootbaH Cuazh VEmployers Unw Asslt roman mam amp Sewices l We yam man aaayaaa Lkag m a We a eyaaaam anaaaam a gt Check um Ees News a Fwd a Jub Januavy vegan New w 7 mamma eeayaaan ana aaa aaaan ama Open yum FREE accuum and pm yum vesume unhne s an Ma ea Seavch 2m WM um FREE amumane JubHumevsW gljggjmftadem Reseavch um dalabase umvev nun 2mg uyevs nu ma wensues39 Laam Maya Get exgen advmz at um new R esuuvce Cemev Access sa avy sJNeysca w amv ve ucauun en s nemmkmg uggmunmea amptvammgesmgmus gt Use thDug cum m seavcmubs quotgm mm yum deskmp Dewmeae Enbvm39 uday E T p aJan sue a FWD nlemel AmWM cahm UM gseanm a w 1 H EIMmmsa PbWevanl Na aa sgavchlxhhemvdymen meg mm mahdmg menu rmm Chns Nhnnmg andhsan ism vvvvv r quot T09 10 u mg i gazeszxvim Andrew M including Example Applications of NLP Automatically Solving Crossword Puzzles I nudeuquot rut I NEWS I OneAcrnss Changes OneAeross is morinr1 to a bigger better server This may cause some problems as we get kinks worked out but hopefully by early next week everything will be in better shape Thanks for your patience I Support this Site re currently provide answers to over lilillllltl searches a day You can help Hello Support this site today P r Click to Pay fully re mdahlg Contact Suort Advertise Privac About I Having trouble getting the last word in that puzzle Having trouble getting the rst1 See if our search engine can help Unlike pure pattern dictionary searches we actually analyze the Clue as well Crossword Clue Search Clue Pattern 5 Go How to Search Enter a clue and either the length of the answer or an answer pattern For unknown letters in the word pattern you ean use a question mark Clue Trout Basket Clue Cut Pattern Pattern n Clue Clue Scheme Pattern aTske Pattern r Hints for better searehine Example Applications of NLP Question Answering I A i A i O 0 O AnswerBus Question Answering System who is married to bill gates 4 i 0 Ahttpwawan5werbuscomlcgi binanswerbusranswero quot 0 microsoft spelling correction 0 Google Slashdut Newsv MCC Researchv Reviewing v Macv Java Thesaurus Fundingv Teachingv LED 7 Yahoo Andrew includin Q AnswerBus Question An who is married to billlr gates Ask Type in your question in English French Spanish ermnn Italian or Portuguese Question who is married to bill gates Possible answers XML m 0 Bill was married to Melinda French Gates in 1994 in Hawaii I Maw Gates Bill s mother biggest fan and strongest nrodder nallv laid down an ultimatum in 993 She was dvine of cancer and warrth to see her only son married I Bill Gates married Melinda French in Hawaii on Januan39 l 1994 and his mother died a few months later t 1994 Bill Gates and Melinda French married in Hawaii on New Years Dav Try your question on other engines Alta Vista CNN News Engine Ask Jeeves Excite Georgie HntBot l chosl Start l Yahoo Example Applications of NLP Machine Translation I WUMSCHZETTEL MEINKDNTD l HILFE v l q EliGLJSH ELEKTRDNIK amp KI39JCHEE commute HOME MEINSHDP BOOKS Hm MSW mm mm SOFTWARE VIDEDSHELE ENDED SPECIALIZED TIME SEARCH I s mEaERN I BESTSELLER I NOVELHes I I BOOKS WRITINGS High speed search German books t Stoebarn All categories GB V K DALLES MUSS RAUS JETZT ZUGREIFEN 5 5Chn pp then 1399quot Solangn dor u39orral micht BUCHINFO Assault or preventive strike The German More to this book aftgalc ggthefSIiEVIet Union on 22 June 1941 1 uses 1050 starting Overview 0 er ar Bum a from EUR Amazonde reader Would you like to sell More of Mstarting from EUR 1050 l Diesen Artikel m Gerhard Baumfalk Offerer dispatches in 12 workingdays What do you mean Their ogl nion to this book Continue to recommend the book b EMa Kategorie Politics Bioqra en amp historv soecialized books Now at money back Now fetch the 1 1 taxes back Paperback 164 sides Rita Fischer Ffm 1 which are Publication date July 1997 ISBN 3895014931 entitled to you AndIechC includingquot with the software and Amazonde Verkaufsrang 204896 Example Applications of NLP Automatically generate Harlequin Romance novels I 1 I 39 LIV H quotLi quotH quot M II inkf39Lt js II E kn 39r quot Andrew McCallum UMass Amherst including material from Chris Manning and Jason Eisner Goals of the Course Introduce you to Natural Language Processing problems and solutions Ultimate focus on handling ambiguity by probabilistic integration of evidence Give you some handson practice with data and a handful of methods mmmmmmmmmmmmmmmmmmmm st This Class Assumes you come with some skills Some basic mathprobability decent programming skills We will use Python tutorial coming next week Some ability to learn missing knowledge Teaches key theory and methods for language modeling tagging parsing etc But it s something like an Al Systems class Hands on with data Often practical issues dominate over theoretical niceties mmmmmmmmmmmmmmmmmmmm st Course Logistics Professor Andrew McCallum TAs David Mimno Karl Schultz Assistants Hanna Wallach Khash Rohanimanesh Time TueThu 230345pm Mailing list 585 staff cs umass edu More information on Web site httpwwwcsumassedumccalumcoursesinp200 Andrew McCallum UMass Amherst including material from Chris Manning and Jason Eisner Grading 7 short written homework programming assignments no way to really internalize without doing it some handson experience should be fun should take about 12 hours each Random informal inclass collaborative quizzes help you set expectations for the midterm and final Final project with a small team mixed backgrounds chance to explore a special interest at end of term Midterm amp Final and classroom participation mmmmmmmmmmmmmmmmmmmm st For Linguistics Students Programming Yipes Yes but with extensive support for those wout expenence Historically popular language for CL courses Prolog clean hard to learn counterintuitive Perl quick but obfuscated syntax messy to read Interpreted rapid prototyping Why Python is bettersuited easy to learn clean syntax powerful features becoming increasingly popular in CompLinguistics Extensive tutorials CompLing support toolkits data etc Many CS students don t know it either put you on more equal footing Andrew McCallum UMass Amherst including material from Chris Manning and Jason Eisner Syllabus Outline Two parts First handson course introductory methods HW Second more like a seminar project First half Language structures and computation Foundation of probability and information theory Use those foundations to work with language Example topics Language models language prediction spam filtering Collocations word clustering word sense disambiguation Finite state machines Markov models Partofspeech tagging Modern parsing techniques Information extraction semantics question answering discourse Andrew McCallum UMass Amherst including material from Chris Manning and Jason Eisner To Do This Week Visit course Web site browse around Read Chapters 1 and 2 in Jurafsky amp Martin textbook Available on line See course web site Install Python on your computer Get extensive help from the TAs if you like mmmmmmmmmmmmmmmmmmmm st Thank you Andrew McCallum UMass Amherst including material from Chris Manning and Jason Eisner
Are you sure you want to buy this material for
You're already Subscribed!
Looks like you've already subscribed to StudySoup, you won't need to purchase another subscription to get this material. To access this material simply click 'View Full Document'