{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "69\n"
     ]
    }
   ],
   "source": [
    "import pandas as pd\n",
    "from nltk.corpus import stopwords\n",
    "from nltk.tokenize import word_tokenize\n",
    "import re\n",
    "\n",
    "musical_plot = pd.read_csv('/Users/김서영/Desktop/datacap/data/musicalData/broadMusicalPlot.csv',encoding='cp949')\n",
    "\n",
    "print(len(musical_plot)) #5699 ==> train 2500 test 2500\n",
    "\n",
    "train_data_size = 1463\n",
    "test_data_size = 1463\n",
    "\n",
    "#전처리(1) 전부 소문자로 변환\n",
    "\n",
    "\n",
    "#romance_plot.줄거리 = romance_plot.줄거리.str.lower()\n",
    "#thriller_plot.줄거리 = thriller_plot.줄거리.str.lower()\n",
    "\n",
    "#전처리(1-1) 데이터 csv 파일로 옮기기\n",
    "#romance_plot 2899개 train_data로 to_csv || 2800개 test_data로 to_csv\n",
    "#thriller_plot 2899개 train_data로 to_csv || 2800개 test_data로 to_csv\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                                               muplot\n",
      "0   This Jason Robert Brown musical features a cas...\n",
      "1   This musical comedy is based on C-R-E-P-U-S-C-...\n",
      "2   This dance-heavy 1980 Tony Award-winner for Be...\n",
      "3   Created by the team behind The Fantasticks 110...\n",
      "4   A musical based on the events surrounding the ...\n",
      "..                                                ...\n",
      "64  A classic musical based on T.H. White's \"The O...\n",
      "65  A Cole Porter classic about the showgirls of t...\n",
      "66  A contemporary operetta based on the novel by ...\n",
      "67  A through-composed musical from Jeanine Tesori...\n",
      "68  The classic Rodgers and Hammerstein musical ba...\n",
      "\n",
      "[69 rows x 1 columns]\n"
     ]
    }
   ],
   "source": [
    "print(musical_plot)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "Mu = [[] for _ in range(69)]\n",
    "for i in range(69):\n",
    "    Mu[i].append(''.join(musical_plot.muplot[i]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[['This Jason Robert Brown musical features a cast and band entirely made up of teenagers.Story:When 13 year old Evan Goldman moves from New York City to Indiana he must escape rumors and blackmail if he hopes to make friends and find happiness at his new school.  This play deals with the hardships of leaving familiar places for unfamiliar ones fitting in with your peers and transitioning from youth to adulthood.'], ['This musical comedy is based on C-R-E-P-U-S-C-U-L-E an original improvisational play created by Rebecca Feldman about a middle school spelling bee.Story:Six spellers - Olive Ostrovsky William Barf챕e Logainne Schwartzandgrubenierre Marcy Park Leaf Coneybear and Chip Tolentino - compete at the annual spelling bee at Putnam Valley Middle School in Putnam County NY.  Flashbacks and insight reveal personal details about each of the spellers in this musical that explores the idea that winning isn?셳 everything.\\t\\tParental Advisory: Contains Explicit Content'], ['This dance-heavy 1980 Tony Award-winner for Best Musical is based on the novel by Bradford Ropes and the 1933 Academy Award-winning musical film.Story:Peggy Sawyer arrives in New York City from Allenstown PA with hopes of becoming a Broadway star.  Facing adversity from all directions the young but talented performer learns about show business and discovers which relationships are most important in life.  With highs and lows in her early career Peggy experiences growth as both a performer and a young woman. Get Tickets for 42nd Street'], ['Created by the team behind The Fantasticks 110 In the Shade is a musical adaptation of the popular play The Rainmaker.Story:Lizzie Curry and her family operate a struggling cattle ranch in a rural drought-ridden town in Texas in the 1930s.  Despite the troubles of the ranch the family?셲 top priority is finding loveless Lizzie a suitable husband. When a charming man named Starbuck arrives in town though Lizzie and her town are turned upside down. '], ['A musical based on the events surrounding the drafting and signing of the Declaration of Independence 1776 won the 1969 Tony Award for Best Musical.Story:John Adams the original innovator behind America?셲 Declaration of Independence teams up with Benjamin Franklin Richard Henry Lee and Thomas Jefferson to battle those in Congress who oppose the movement to declare independence from England.  The team cooperatively conceives and eventually passes the resolution that frees America from British tyranny. Fun FactHoward Da Silva had a heart attack after the final preview of 1776 but insisted on performing on opening night.\\t\\t\\t\\t\\t\\t\\tShow & Tell: The New Book of Broadway Anecdotes by Ken Bloom'], ['This musical comedy is based on the Addams Family characters featured in Charles Addams??single-panel gag cartoons.Story:Wednesday Addams and her ?쐍ormal??boyfriend Lucas arrange a dinner at the Addams??mansion to announce to their families their plans to marry.  The idea is jarring to both the families and when the entire groups winds up locked in overnight secrets come out and revelations abound. '], [\"A musical revue celebrating Duke Ellington's years at Harlem's Cotton Club.Story:After Midnight takes the sexy smoky glamour of the original Jazz Age and catapults it into a whole new era of heart-pounding mind-blowing entertainment for modern Broadway audiences. Refracted through a contemporary lens After Midnight celebrates Duke Ellington's years at the Harlem's famed Cotton Club using his original arrangements and performed by a world-class big band of 17 musicians hand-picked by living jazz legend Wynton Marsalis. The timeless tunes set against a narrative of Langston Hughes poetry provides an authentic backdrop for an array of cutting-edge performances by 25 sensational vocalists and dancers including special guest stars whose interpretations will shatter everything you think you know about music nightlife and Broadway.\"], ['Featuring music by Elton John and lyrics by Tim Rice this Disney-produced musical is based on Giuseppe Verdi?셲 19th-century opera of the same name.Story:Conflict between Egypt and Nubia gives birth to a star-crossed love affair between Aida the captured princess of Nubia and Radames the Egyptian captain who enslaved her people.  With love and responsibility at odds Aida and Radames must decide to follow their hearts or lead their people.  '], [\"Named after the 1929 Fats Waller song Ain?셳 Misbehavin??is a musical revue tribute to the black musicians of the Harlem Renaissance in the 1920s and '30s.\"], [\"A new jukebox musical featuring the story and music of the Temptations.Story:Ain't Too Proud follows The Temptations??journey from the streets of Detroit to the Rock & Roll Hall of Fame. With their signature dance moves and unmistakable harmonies they rose to the top of the charts creating an amazing 42 Top Ten Hits with 14 reaching number one. Through friendship and betrayal amid the civil unrest that tore America apart their moving and personal story still resonates five decades later.\"], ['This Disney-produced musical comedy is based on the original animated classic.Story:Poor hooligan Aladdin has big aspirations -- he dreams of a life in the palace with his beloved Princess Jasmine.  When he chances upon a lamp that holds an entrapped genie Aladdin is allowed to experience the life he?셲 always imagined.  As he matures from boy to man Aladdin is forced into decisions that will affect both his future and the futures of others.'], ['In this musical revue Michael Feinstein and Dame Edna go head to head in a mashup of classic American songs and bright brash comedy.'], ['Inspired by the early life experiences of actor George Takei Allegiance tells one family?셲 unheard story of intense patriotism during the Japanese American internment of World War II.Story:Sammy and Kei Kimura Japanese American siblings living in California hold opposing views of how best to counter the anti-Japanese sentiments sweeping America after the attack on Pearl Harbor.  Sammy elects to join the 442nd Regimental Brigade as a sign of loyalty and solidarity while Kei protests the unjust treatment of her race and calls for fairness.  In a time of chaos and fear both protagonists seek freedom and acceptance for themselves their family and their culture. '], ['The third musical created by the legendary duo Rodgers and Hammerstein Allegro is a serious work that focuses on the everyday struggles of the ordinary man.Story:?쏣veryman??Joseph Taylor Jr. (Joe) is born to a middle-class family in small town America.  The son of a doctor and an intelligent student with a promising future Joe marries the love of his youth and accepts a high-profile position at a Chicago hospital to support himself and his wife Jennie during the Depression.  Hardships with Jennie and unhappiness with big city life help Joe realize his missteps in life and spur his move back home to become a small-town doctor. '], ['Altar Boyz is a musical comedy that satirizes the popularity of boy bands and Christian-themed music in contemporary American culture. Story:The Altar Boyz perform the final concert of their ?쏳aise the Praise??tour in real time.  The five performers named Matthew Mark Luke Juan and Abraham perform their popular and entertaining pieces and reveal tidbits about themselves and the band throughout the night. The boys employ a funny machine to detect burdened souls in the performance space and reveal the goal to bring the ticker to zero by the end of the show. '], ['This musical conveys the emotional story behind the revered Christian hymn and the man who penned it - Englishman John Newton.Story:Englishman John Newton looks forward into an uncertain future - on one hand feeling destined to join his father in the ranks of slave traders on the other feeling personally inspired to embrace the more compassionate views of his childhood sweetheart.  A violent storm at sea terrifies John and becomes the inspiration for the song which will become an anthem for people around the world for centuries to come - and change the life of the man himself.'], ['A new musical based on the French film of the same name.Story:Am챕lie follows the journey of the inquisitive and shy Am챕lie who turns the streets of Montmartre into a world of her own imagining while secretly orchestrating moments of joy for those around her. After discovering a mysterious photo album and meeting a handsome stranger Am챕lie realizes that helping others is easier than participating in a romantic story of her own.\\t\\t\\t(adsbygoogle = window.adsbygoogle || []).push({});\\t\\t\\t'], ['This romantic dance musical features songs of Ira and George Gershwin and is based on the 1951 Academy Award-winning film starring Gene Kelly.Story:After World War II American veteran Jerry Mulligan chooses to stay in Paris to pursue a career as a painter.  When he and the lovely and unavailable French girl Lise Bouvier fall for each other both of their lives become complicated.'], ['This sung-through jukebox musical is an adaptation of the 2004 rock opera of the same name originally written and recorded by punk band Green Day. Story:Three American youths Johnny Tunny and Will seek to escape the media-saturated life of suburbia.  The boys strike out in various directions with Johnny seeking solace in the city Tunny enlisting in the Army and Will staying home with his pregnant girlfriend.  The individual journeys challenge the boys with each facing considerable hardships and finding few rewards.  Eventually they all return home with hard-won maturity and a new outlook on life. '], ['A musical version of the novel by Bret Easton Ellis and the film starring Christian BaleStory:American Psycho follows 26-year-old Patrick Bateman: a sophisticated rich and devastatingly handsome Wall Street banker in 1980s New York City. He?셲 got a sculpted body a model-gorgeous girlfriend and a to-die-for apartment. There?셲 just one snag: Patrick can?셳 get the blood out of his $5000 suits because he also has a murderous psychopathic alter ego that he hides from his friends and co-workers.Parental Advisory: Contains Explicit Content'], ['The stage adaptation of the 1997 film featuring a score by Ahrens and Flaherty.Story:From the twilight of the Russian Empire to the euphoria of Paris in the 1920s the new musical is the story of a brave young woman attempting to discover the mystery of her past while finding a place for herself in the rapidly changing world of a new century.'], ['This family-friendly 1977 Tony Award-winner for Best Musical is based on Harold Gray?셲 popular comic strip ?쏬ittle Orphan Annie??which he started in the 1920s.Story:Little orphan Annie is taken from her orphanage and the clutches of evil Miss Hannigan to stay at billionaire Oliver Warbuck?셲 mansion for the holidays. When Annie wins the hearts of Warbucks and his staff they set out on an epic mission to find Annie?셲 parents.  Meanwhile Miss Hannigan in partnership with her brother Rooster and his girlfriend Lily attempt to make some money out of this unique situation.  The stories become intertwined as the lovable Annie?셲 story hits the national spotlight.'], [\"A classic musical about the fictionalized life of sharpshooter Annie Oakley from songwriter Irving Berlin.Story:Annie Oakley and Frank Butler meet when she beats him in a sharpshooting contest at Buffalo Bill's Wild West Show and the two have to negotiate between their completive sides and their blossoming love for each other.\"], ['A cult favorite this Sondheim musical only played 9 performances in its initial run.Story:When Cora Hoover Hooper the hated mayor of a small town is faced with a town bankruptcy she concocts a fake \"miracle\" to bring in tourism. Unfortunately during the tourist rush the inmates from the local asylum the Cookie Jar escape and the lines between sanity and insanity quickly become blurred. '], ['Originally penned in 1934 with music and lyrics by American composer and songwriter Cole Porter this musical comedy has been updated several times throughout the last century.Story:Billy instantly falls in love with a beautiful girl he meets in a taxi.  When he discovers she?셲 boarding the SS American the same London-bound ship his boss and co-worker Reno are boarding he sneaks aboard himself.  The beautiful girl Hope is engaged to a stuffy British aristocrat Lord Evelyn but that doesn?셳 stop the love-struck Billy.  With the help of other passengers Billy seeks to shake Reno whose love he doesn?셳 return and capture the heart of the girl of his dreams - all without hurting anyone?셲 feelings.'], ['The musical about backstage backstabbing based on the film All About Eve and Mary Orr\\'s \"The Wisdom of Eve\"Story:Eve is Margo Channing\\'s biggest fan but isn\\'t happy until she is able to emulate her in every single way - stooping to steal her friends fiance and standing as Broadway\\'s biggest star. '], [\"Three acts containing stories based off of Mark Twain's The Diary of Adam and Eve Frank R. Stockton's The Lady or the Tiger? and Jules Feiffer's Passionella.\"], ['The Andrew Lloyd Webber musical based on the novella of the same name by David Garnett.Story:Actress Rose Vibert is having an affair with her biggest fan Alex when she meets his uncle George who though in love with her is still entangled with his mistress Giulietta Trapani. George and Rose marry to have a daughter named Jenny who falls in love with Alex who himself is now infatuated with Giulietta. As Alex reflects \"love changes everything.\"'], ['A Sondheim musical tracing Leon Czolgosz John Hinckley Charles Guiteau Giuseppe Zangara Samuel Byck Lynette \"Squeaky\" Fromme Sara Jane Moore and John Wilkes Booth\\'s attempts at assassinating the President of the United States of America.'], ['This 2004 Tony Award짯winner for Best Musical features colorful ?쏶esame Street?씲춖tyle puppets with real짯life problems and adult vocabularies.Story:Princeton a recent college graduate with a ?쏝.A. in English??finally finds an apartment he can afford 짯 on Avenue Q. Wondering about his purpose in life Princeton takes advice from his eccentric neighbors. At the same time he encounters and begins to fall in love with Kate an aspiring teacher but struggles with his own fear of commitment. During Princeton?셲 adventures with his new Avenue Q friends he discovers the true meaning of friendship and develops his own sense of purpose.Parental Advisory: Contains Explicit Content'], [\"The classic 'Hey kids let's put on a show!' musical featuring tons of recognizable songs.Story:When a group of teenagers' parents leave to join the dying art of vaudeville the teens stage a revue to keep from being sent to a work farm. The first show is a bomb but when a French pilot lands nearby they are able to get enough publicity to put on a great show and start their own youth center.Apply for Guare Version Production RightsApply for Oppenheimer Version Production Rights\"], ['Based on a story developed with Susan Yankowitz about three expecting couples.Story:Three couples each newly expecting a child have different but familiar reactions. Lizzie and Danny are university juniors who have just moved in together. Athletic Pam and her husband Nick a sports instructor have had some trouble conceiving. Arlene already the mother of three grown daughters is unsure of what to do contemplating abortion while her husband Alan is thrilled with the thought of a new baby. Throughout the show these characters experience the emotional stresses and triumphs the desperate lows and the comic highs that accompany the anticipation and arrival of a baby.'], [\"A new jukebox musical about Florence Greenberg the New Jersey housewife who helped the pop group The Shirelles.Story:When a group of African-American girls performing at a talent contest come to Florence Greenberg's attention she jumps at the opportunity to make them recording artists and founds Scepter Records. She then goes on to discover  like The Kingsmen The Isley Brothers Dionne Warwick and others.\"], ['A David Yazbek musical based on the screenplay by Eran Kolirin celebrating the deeply human ways music and laughter connect us all.Story:An Egyptian Police Band arrives in Israel to play a concert. After a mix-up at the border they are sent to a remote village in the middle of the desert. With no bus until morning and no hotel in sight these unlikely travelers are taken in by the locals. Under the spell of the desert sky their lives become intertwined in the most unexpected ways. \\t\\t\\t(adsbygoogle = window.adsbygoogle || []).push({});\\t\\t\\t'], ['Laura Osnes and Corey Cott star in this new musical set amidst the 1940s swing scene.Story:Set in 1945 Bandstand tells the story of musician Donny Novitski who is about to take on the mission of his life: leading his band of fellow veterans into competition for America?셲 next swing band sensation. When Donny meets a beautiful young singer named Julia he finds the perfect harmony in words and music that could take this band of brothers all the way to the live radio broadcast finale in New York City. But to succeed it will require every ounce of talent stamina and raw nerve that these musicians possess.'], [\"A musical based on the life of showman P. T. Barnum.Story:PT Barnum is in the circus business despite his wife's wishes working with talents such as the oldest woman alive Joice Heth the tiny Tom Thumb and the famous Swedish opera singer Jenny Lind with whom he falls in love. In the end he returns to his wife and joins with another circus owner James Anthony Bailey to form the Barnum and Bailey circus.\"], ['The musical about a pill that helps losers become popular. Based on the novel by Ned Vizzini.Story:Achieving the ?쐏erfect life??is now possible thanks to some mysterious new technology - but it comes at a cost that?셲 not as easy to swallow. What could possibly go wrong? Blending the contemporary with retro sci-fi this thrillingly exciting comically subversive and deeply felt new musical takes on the competing voices in all of our heads. And ultimately proves there?셲 never been a better time in history to be yourself?봢specially if you?셱e a loser?쫎eek?쫛r whatever.'], ['This biographical musical tells Carole King?셲 story and features songs from her legendary music career.Story:Carole King (then named Carol Klein) starts her music career as a 16-year old songwriter selling originals in Manhattan but when she meets her future husband and music partner Gerry Goffin her career begins to blossom.  Through relationships both inside and outside the studio Carole finds her unique musical voice and finally records the solo album that will make her a star.'], ['This first of Disney?셲 stage adaptations is based on the on the 1991 animated Disney film which was adapted from the French fairy tale by Jeanne-Marie Leprince de Beaumont.Story:A wandering enchantress transforms a cruel and vain prince into a hideous Beast leaving him only one way to reverse the curse - fall in love with another and earn her love in return.  Belle a beautiful book-lover encounters the Beast when she arrives at his castle to plead for her father?셲 freedom ultimately trading her own for his.  As tension over her imprisonment escalates in the town spurred by Gaston the selfish lover who seeks Belle?셲 hand in marriage the relationship between Belle and the Beast grows leading to an emotional and transformative conflict.'], ['An original musical based on the Tim Burton film of the same name.NOTE: Includes explicit language.Story:The story of Lydia Deetz a strange and unusual teenager obsessed with the whole ?쐀eing dead thing.??Lucky for Lydia her new house is haunted by a recently deceased couple and a degenerate demon who happens to have a thing for stripes. When Lydia calls on this ghost-with-the-most to scare away her insufferable parents Beetlejuice comes up with the perfect plan which involves exorcism arranged marriages and an adorable girl scout who gets scared out of her wits.'], [\"A classic musical based on Mary Printz who worked for Green's answering service. Later adapted into a film of the same name.Story:Telephone operator Ella Peterson adopts different personalities while on the job. Her life is perfect until she begins a complicated romance with one of her clients Jeff Moss and the police start looking in on her boss's operations.\"], ['Based on a story by Larry L. King that was inspired by the real-life Chicken Ranch in La Grange Texas.Story:It is the late 1970s and a brothel has been operating outside of fictional Gilbert Texas for more than a century. It is under the proprietorship of Miss Mona Stangley who is on good terms with the local sheriff Ed Earl Dodd. When crusading television reporter Melvin P. Thorpe decides to make the illegal activity an issue political ramifications cause the place to be closed down.'], [\"A musical based on the hit film about a boy who dreams of being big.Story:Josh Boskin makes a wish to become big. When his wish comes true he lands a dream job working for Maximillion toys and falls in love Susan. But in the end he must decide whether he's ready to live a grown up's life yet.\"], [\"A new musical based on the novel by Daniel Wallace and the award-winning Tim Burton film.Story:A rollicking fantasy set in the American South Big Fish centers on the charismatic Edward Bloom whose impossible stories of his life's epic adventures frustrate his son Will. Now with Edward on his deathbed Will must embark on his own journey to find out who his father really is unraveling the man from the myth the truth from the tall tales.\"], ['A new musical based on Mark Twain\\'s novel Adventures of Huckleberry Finn.Story:Though trying to stay \"respectable\" Huckleberry Finn runs into trouble when his father Pap returns to town. Huck is forced to fake his own death to escape and joins up with another on the run - Jim a slave in the search for freedom. As they head down the river they get sucked into scams involving two \"aristocrats\" one of which lands Jim in chains once again and Huck and Tom Sawyer join together to free him.'], ['This 2009 Tony Award-winner for Best Musical set during the 1984-85 miners??strike in the United Kingdom is based on the 2000 film and features music by Elton John.Story:Billy is a motherless eleven-year-old boy in northeast England during the miners??strike of 1984-85 - a difficult time for miners like his father and brother.  Attracted to the grace of ballet Billy leaves his boxing class to begin taking dance lessons. When his father finds out and forbids them he begins taking secret private lessons with Mrs. Wilkinson who believes he has the talent to be accepted into the Royal Ballet School in London.  As Billy?셲 hopes and dreams grow bigger and the strike escalates the family and the community are both challenged to consider and fight for that which is truly most important.'], ['A contemporary musical based loosely on the 1844 novella The Corsican Brothers by Alexandre Dumas the West End production of this musical is one of the longest-running musical productions in history.Story:Mrs Johnstone is working as a cleaner for Mrs Lyons when she becomes pregnant with twins. The financial burden of two children is too much for her and Mrs Lyons is longing for a child of her own so Mrs Johnstone keeps Mickey and gives the other boy Edward to Mrs Lyons. Though the blood brothers are never told they are related and their mothers try their best to keep them apart their paths do cross later in life and all hell breaks loose. '], ['A new rock musical tracing the rise of Andrew Jackson from early childhood through to his controversial presidency.'], ['The new Bollywood-style musical transfer from the UK.Story:Akaash is a young man from the slums of Bombay who dreams of becoming the next big star in Bollywood. Priya the daughter of a famous Bollywood director saves his slum helps him rise is show business and steals his heart but she is engaged to another and from an entirely different class.'], [\"A hip-hop version of Shakespeare's A Comedy of Errors.Story:Identical twins both named Antipholus are served by another set of identical twins both named Dromio. The twins separated since birth arrive in the same city and a wonderful series of mishaps occur as the twins are mistaken for each other.\"], ['Based on the real-life story of Bonnie Elizabeth Parker and Clyde Chestnut Barrow outlaws robbers and criminals who traveled the United States during the Great Depression.Story:Bonnie has dreams of a life in the movies when she meets Clyde who has just broken out of prison with his brother Buck. The two team up and begin a life of crime to become the stuff of folk legend. '], ['Matt Stone and Trey Parker the creators of TV?셲 South Park collaborate with Avenue Q co-composer/co-lyricist Robert Lopez on this irreverent musical comedy.Story:Elder Price and Elder Cunningham evangelists of the The Church of Jesus Christ of Latter-day Saints are assigned to spread Mormonism in Uganda.  Though they struggle with cultural differences their own social oddities and their inadequate knowledge of actual Mormonism they commit to helping the Ugandans find the road to salvation - whatever that means. Parental Advisory: Contains Explicit Content'], ['A new musical starring Hugh Jackman based on the life of singer Peter Allen a man who began in the outback and rose to fame under the supervision of Judy Garland and the love of her daughter Liza Minella.\\t\\t\\t(adsbygoogle = window.adsbygoogle || []).push({});\\t\\t\\t'], [\"The Rodgers and Hart musical adaptation of Shakespeare's A Comedy of Errors.Story:Identical twins both named Antipholus are served by another set of identical twins both named Dromio. The twins separated since birth arrive in the same city and a wonderful series of mishaps occur as the twins are mistaken for each other.\"], [\"Jason Robert Brown and Marsha Norman's adaptation of James Waller's novel of the same name.Story:The Bridges of Madison County tells the story of photographer Robert Kincaid and his life-changing four-day love affair with Iowa farm wife Francesca Johnson.\"], ['The musical of a classic Scottish story and a love that transcends time.Story:Two American tourists Tommy Albright and Jeff Douglas stumble upon a misty little Scottish town that only awakens once every hundred years.'], [\"Steve Martin and Edie Brickell's new musical set in the Blue Ridge Mountains.Story:Featuring a score by Steve Martin and Edie Brickell Bright Star is set in the Blue Ridge Mountains of North Carolina and travels between 1945 and 1923. Billy Cane a young soldier just home from World War II meets Alice Murphy the brilliant editor of a southern literary journal. Together they discover a powerful secret that alters their lives.\"], ['Based loosely on the film of the same name a new musical from the creators of In the Heights and Next to Normal.Story:When school redistricting sends popular cheer-captain Campbell to Jackson High School she is suddenly on the outs popularity-wise at the new multi-ethnic school. In order to get back on top Campbell has to convince Danielle the head of Jackson\\'s hip-hop dance troop to start a cheer squad if she\\'s ever going to \"bring it on\" and end up on top again.'], ['A streetwise new musical based on the real life story of Chazz Palminteri and the play and film of the same name.Story:A Bronx Tale is set against the backdrop of racial strife and organized crime in the 1960s. It is the story of an Italian-American teenager finding his path in life as he must choose between the father who raised him and a mob-boss father figure who fascinates him.\\t\\t\\t(adsbygoogle = window.adsbygoogle || []).push({});\\t\\t\\t'], [\"A new musical adaptation of Woody Allen and Douglas McGrath's comedy.Story:Bullets Over Broadway tells the story of an aspiring young playwright who is forced to cast a mobster's talentless girlfriend in his latest drama in order to get it produced. Featuring music of the 1920s the musical includes a new book by Woody Allen.\"], ['A ballroom dance musical featuring Tangos Sambas Mambos Quicksteps Swing Dancing Viennese Waltzes Jives and Paso Dobles.'], ['A musical inspired by the phenomenon of Elvis Presley and his draft notice into the Army from composer Charles Strouse.Story:Conrad Birdie has been drafted to the army. As a farewell the rock and roll star is scheduled to sing Albert Peterson\\'s \"One Last Kiss\" and kiss a girl from Sweet Apple Ohio live on the Ed Sullivan Show. However Conrad\\'s hip-thrusting makes more than one man in town uncomfortable many of whom attempt to stop the show from happening. Meanwhile Rosie wonders if Alfred for whom she does everything will ever leave his mother\\'s shadow.'], ['Based on a 1939 novel by Christopher Isherwood and a 1951 play by John Van Druten this musical features music by John Kander and lyrics by Fred Ebb.Story:Cliff Bradshaw an American author working on a novel in Berlin encounters Sally Bowles at the seedy Kit Kat Klub where she sings.  When she is fired by the club?셲 owner also her jealous boyfriend she moves in with Cliff and the two fall in love.  As the Nazis begin taking control of the German government the atmosphere of the Kit Kat Klub and the lives of Cliff and Sally begin to change dramatically.'], ['A political-satire star-vehicle for Ethel Merman by Irving Berlin.Story:Sally Adams becomes Ambassador of Lichtenburg and charms Cosmo Constanstine with her undiplomatic manner. Meanwhile her press attache Keneth Gibson falls in love with Princess Maria.'], ['A classic musical based on T.H. White\\'s \"The Once and Future King.\"Story:Guinevere King Arthur\\'s lovely wife encourages him to establish the Knights of the Round Table. As the Knights grow in fame Lancelot Du Lac becomes the most celebrated of all of them and rumors begin to circulate about his affair with Guinevere. Can Camelot be the idealistic place of Arthur\\'s dreams?'], ['A Cole Porter classic about the showgirls of the Montmartre dance halls during the 1890s.Story:La M척me Pistache decides to feature the scandalous Can-Can dance in her dance hall but Judge Forestier threatens to close down her establishment if she defies the law.'], ['A contemporary operetta based on the novel by Voltaire.Story:Candide and his friends Cunegonde Paquette Maximilian Dr. Pangloss and the old lady (with one buttox) survive numerous disasters but never give up their search for truth in this \"best of all possible worlds.\"Apply for 1973 Version Production RightsApply for 1999 Version Production Rights'], [\"A through-composed musical from Jeanine Tesori and Tony Kushner the author of Angles in America.Story:Caroline is a maid in the Gellman household and both Caroline's family and the Gellmans are struggling. Noah and Caroline form a unique relationship as each attempts to deal with the change in their lives.\"], ['The classic Rodgers and Hammerstein musical based on a play by Ferenc Molnar.Story:Billy Bigelow a traveling carnival man and Julie Jordan a local factory worker are going to have a child. Desperate for money Billy attempts to rob a bank and is killed.']]\n"
     ]
    }
   ],
   "source": [
    "print(Mu)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|█████████████████████████████████████████████████████████████████████████████████| 69/69 [00:00<00:00, 633.69it/s]\n"
     ]
    }
   ],
   "source": [
    "# 토큰화+전처리(3) 전체 불용어 처리\n",
    "# 전체 플롯\n",
    "from tqdm import tqdm\n",
    "Mu_vocab = {} \n",
    "Mu_sentences = []\n",
    "stop_words = set(stopwords.words('english'))\n",
    "\n",
    "for i in tqdm(Mu):\n",
    "    Mu_sentences = word_tokenize(str(i)) # 단어 토큰화를 수행합니다.\n",
    "    result = []\n",
    "    for word in Mu_sentences: \n",
    "        word = word.lower() # 모든 단어를 소문자화하여 단어의 개수를 줄입니다.\n",
    "        if word not in stop_words: # 단어 토큰화 된 결과에 대해서 불용어를 제거합니다.\n",
    "            if len(word) > 2: # 단어 길이가 2이하인 경우에 대하여 추가로 단어를 제거합니다.\n",
    "                result.append(word)\n",
    "                if word not in Mu_vocab:\n",
    "                    Mu_vocab[word] = 0 \n",
    "                Mu_vocab[word] += 1\n",
    "    Mu_sentences.append(result) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "Mu_vocab_sorted = sorted(Mu_vocab.items(), key = lambda x:x[1], reverse = True)\n",
    "\n",
    "#전처리(4) 인덱스 부여\n",
    "Mu_word_to_index = {}\n",
    "i=0\n",
    "for (word, frequency) in Mu_vocab_sorted :\n",
    "    if frequency > 1 : # 정제(Cleaning) 챕터에서 언급했듯이 빈도수가 적은 단어는 제외한다.\n",
    "        i=i+1\n",
    "        Mu_word_to_index[word] = i\n",
    "#print(all_word_to_index)\n",
    "\n",
    "'''vocab_size = 5000 #상위 15000개 단어만 사용\n",
    "words_frequency = [w for w,c in Mu_word_to_index.items() if c >= vocab_size + 1] # 인덱스가 200 초과인 단어 제거\n",
    "for w in words_frequency:\n",
    "    del Mu_word_to_index[w] # 해당 단어에 대한 인덱스 정보를 삭제\n",
    "\n",
    "    \n",
    "Mu_word_to_index['OOV'] = len(Mu_word_to_index) + 1\n",
    "'''\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}