[ Skip to the content ]

Institute of Formal and Applied Linguistics Wiki

[ Back to the navigation ]

You are here: start » user » hladka » playcoref

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision 		Previous revision
Next revision Both sides next revision
user:hladka:playcoref [2009/02/25 21:22]
hladka 		user:hladka:playcoref [2009/02/25 22:24]
hladka
Line 10: Line 10:
  
 ====== Specification ====== ====== Specification ======
 +
 +
 +
  
  
Line 23: Line 26:
   * A game of two players. Players are paired randomly. Computer as a player: automatic coreference resolution **???????**   * A game of two players. Players are paired randomly. Computer as a player: automatic coreference resolution **???????**
   * Session time up to **???????** minutes.   * Session time up to **???????** minutes.
-  * At the beginning, two first sentences of the document are displayed to each player. The players hook up the nouns and pronouns which refer to the same object independently of each other. If a player hooks up all the related words (s)he keeps in mind then (s)he asks for the next sentence of the document. The session goes on until the end of the session time. (//vypustila jsem tu variantu, ze rychlejsi hrac muzi partii ukoncit kdykoli. Jednalo by se vlastne o znevyhodneni pomalejsiho hrace.//) The player who asks for more sentences in the session obtains bonus points.+  * At the beginning, two first sentences of the document are displayed to each player. The players hook up the nouns and pronouns which refer to the same object independently of each other. If a player hooks up all the related words in the given sentences (s)he keeps in mind then (s)he asks for the next sentence of the document. The session goes on this way until the end of the session time. (//vypustila jsem tu variantu, ze rychlejsi hrac muzi partii ukoncit kdykoli. Jednalo by se vlastne o znevyhodneni pomalejsiho hrace.//) The player who has asked for more sentences in the session obtains bonus points.
    * What my partner is doing? If (s)he hooks up the same pair of words as hooked up then the pair of words starts **???????**. If (s)he links a word I have not linked so far then a given word starts **???????**    * What my partner is doing? If (s)he hooks up the same pair of words as hooked up then the pair of words starts **???????**. If (s)he links a word I have not linked so far then a given word starts **???????**
    * The players can re-hook up any word any time in the session.       * The players can re-hook up any word any time in the session.   
 +   * To design the game for a particular language the following data and tools are needed (or are welcome):
 +     - corpus of manually anotated coreference
 +     - POS tagger
 +     - coreference resolution procedure
 +
 +
 +
 +
  
  
Line 35: Line 46:
  
 === Text Selection === === Text Selection ===
-To design the game for a particular language the following data and tools are needed (or are welcome): 
-  - corpus of manually anotated coreference 
-  - POS tagger 
-  - coreference resolution procedure 
- 
   * CS data ^JM^   * CS data ^JM^
      * Anja's data    ## // PDT data that are currently being annotated for the extended coreference //      * Anja's data    ## // PDT data that are currently being annotated for the extended coreference //
Line 49: Line 55:
  
 === Internal format === === Internal format ===
-  * sgml ## //propose dtd file: include the element ''cor'' and its attribute ''src'' (manully, automatically, playes) into csts.dtd// +  * sgml ## //propose dtd file: include the element ''cor'' and its attribute ''src'' (manually, automatically, players) into csts.dtd; ve zkratce: u kazde slovni formy <f_id> bude povinne <l>, <t> a volitelne <cor src="man">, <cor src="Linh">, <cor src="player_id">, ..., <cor src="player_id">. Kdyz se bude s dokumentem hrat poprve, tak tam <cor src="player_id"> nebude zadne//
-  * conversion: csts <-> pml m_coref scheme+
  
 === (Pre)processing === === (Pre)processing ===
-  * tagging +  * tagging ## //see Tools needed below// 
-  * t-parser by Linh+  * acr by Linh ## // dtto //
  
 === Text handling === === Text handling ===
    * sentence by sentence    * sentence by sentence
    * supervised selection of documents for a session     * supervised selection of documents for a session 
 +
 +
 +
 +
  
 ===== Scoring ===== ===== Scoring =====
-  * top lists +   * ''pts_of_player_A = w1*(player_A's_output vs. manual_annotation(if available)) + w2*(player_A's_output vs. automatic_annotation) + w3*(player_A's_output vs. player_B's_output) + speed_pts'' 
 +  
 +// w1 by mela byt nejvyssi; w2 by mela urcite nejak zohlednit uspesnost automaticke procedury - uspesnost merenou na jakych datech?; w3: kdyz hracum budeme zobrazovat i ta slova, ktera oznacil protihrac, a ja je neoznacila, nebudeme je tim tlacit do vynucene shody? pro to, aby w3 bylo 'rozumne', bych zobrazovala pouze dvojice, na kterych se oba hraci shodli.//
  
  
-===== Output Data =====+ 
 +===== Output Data Needed ===== 
 +   * score list ## //player_id, pts, #sessions// 
 +   * documents after the ''n''-th session consists of ''2*n'' players coreference annotation (some of them should be identical, the more identical the better) 
 +   * session 
 +      * player_A_id, player_B_id 
 +      * document(s) 
 +      * number of corrections by player_A and by player_B 
 +      * corrections by player_A and by player_B 
  
 ===== Design ===== ===== Design =====
 +   * What info to be displayed in the session?
 +      * session time = elapsed time + remaining time
 +      * how many sentences my partner has read so far 
 +      * running pts **???????**
 +   * Visualization of the coreference pairs
 +      * colors
 +      * arrows
 +      * ...
  
  
Line 75: Line 103:
    * tagger ^BH^ ## tool_chain (CAC2.0)    * tagger ^BH^ ## tool_chain (CAC2.0)
    * Linh's coreference resolution procedure ^PS^ ## What type of input data the Linh's procedure works with? ''tool_chain'' is going to be extended by the ''S'' option enabling to run Vasek Klimes' t-parser in a basic version, i.e. just t-tree and functors.     * Linh's coreference resolution procedure ^PS^ ## What type of input data the Linh's procedure works with? ''tool_chain'' is going to be extended by the ''S'' option enabling to run Vasek Klimes' t-parser in a basic version, i.e. just t-tree and functors. 
 +  * conversion: csts <-> pml m_coref scheme
  
 ====== Motivační publikace ====== ====== Motivační publikace ======
Line 84: Line 113:
    * Luis von Ahn, Laura Dabish: //Labelling Images with a Computer Game//, 2004,  ([[http://ufal.mff.cuni.cz/~hladka/PlayCoref/ESP.pdf|pdf]]) ## nálepkování obrázků hrou ESP    * Luis von Ahn, Laura Dabish: //Labelling Images with a Computer Game//, 2004,  ([[http://ufal.mff.cuni.cz/~hladka/PlayCoref/ESP.pdf|pdf]]) ## nálepkování obrázků hrou ESP
  
-===== Anotování koreference v českých datech =====+ 
 +====== Anotování koreference v českých datech ======
   * PDT 2.0 [[http://ufal.mff.cuni.cz/pdt2.0/doc/manuals/cz/t-layer/html/ch08.html|odkaz do tektogramatického manuálu]]   * PDT 2.0 [[http://ufal.mff.cuni.cz/pdt2.0/doc/manuals/cz/t-layer/html/ch08.html|odkaz do tektogramatického manuálu]]
   * rozšířená koreference - viz přehled [[https://wiki.ufal.ms.mff.cuni.cz/user:hladka:data]]   * rozšířená koreference - viz přehled [[https://wiki.ufal.ms.mff.cuni.cz/user:hladka:data]]
Line 91: Line 121:
  
  
-===== Automatické určování koreference v českých datech - přehled =====+ 
 + 
 +====== Automatické určování koreference v českých datech - přehled ====== 
   * Dosavadní experimenty   * Dosavadní experimenty
  
Line 100: Line 133:
  
  
-===== Návrh hry - brainstorming =====+ 
 +====== Návrh hry - brainstorming ====== 
 **26/5/08 Anja, Bára:** **26/5/08 Anja, Bára:**
    * Vstup: Texty v povrchové podobě, tedy NE tektogramatické stromy    * Vstup: Texty v povrchové podobě, tedy NE tektogramatické stromy
Line 128: Line 163:
       * **Vystup:** dvojice, na kterych se hraci shodli a neshodli. Shoda bude prinosna, ale co delat s temi dvojicemi, ktere oznacil pouze jeden hrac? Muze se objevit jeste jedna sance na shodu, a to ta, ktera prijde z jine partie se stejnym dokumentem. Proto bude muset byt kladen duraz na rizene podsouvani dokumentu do partii.       * **Vystup:** dvojice, na kterych se hraci shodli a neshodli. Shoda bude prinosna, ale co delat s temi dvojicemi, ktere oznacil pouze jeden hrac? Muze se objevit jeste jedna sance na shodu, a to ta, ktera prijde z jine partie se stejnym dokumentem. Proto bude muset byt kladen duraz na rizene podsouvani dokumentu do partii.
       * Kontrolni mechanismy, aby hraci nespojovali libovolna dve slova - vzhledem k navrhu na pocitani skore, tj. za shodu, budou hraci aspon castecne motivovani hrat smysluplne.        * Kontrolni mechanismy, aby hraci nespojovali libovolna dve slova - vzhledem k navrhu na pocitani skore, tj. za shodu, budou hraci aspon castecne motivovani hrat smysluplne. 
 +
 +
 +
 +
 +====== Specification ======
 +
 +
 +
 +
 +
 +
 +
 +
 +
 +
 +
 +
 +===== Strategy =====
 +  * **Hook up the words which refer to the same entity.**
 +  * A game of two players. Players are paired randomly. Computer as a player: automatic coreference resolution **???????**
 +  * Session time up to **???????** minutes.
 +  * At the beginning, two first sentences of the document are displayed to each player. The players hook up the nouns and pronouns which refer to the same object independently of each other. If a player hooks up all the related words in the given sentences (s)he keeps in mind then (s)he asks for the next sentence of the document. The session goes on this way until the end of the session time. (//vypustila jsem tu variantu, ze rychlejsi hrac muzi partii ukoncit kdykoli. Jednalo by se vlastne o znevyhodneni pomalejsiho hrace.//) The player who has asked for more sentences in the session obtains bonus points.
 +   * What my partner is doing? If (s)he hooks up the same pair of words as hooked up then the pair of words starts **???????**. If (s)he links a word I have not linked so far then a given word starts **???????**
 +   * The players can re-hook up any word any time in the session.   
 +   * To design the game for a particular language the following data and tools are needed (or are welcome):
 +     - corpus of manually anotated coreference
 +     - POS tagger
 +     - coreference resolution procedure
 +
 +
 +
 +
 +
 +
 +
 +
 +
 +
 +===== Input Texts =====
 +
 +=== Text Selection ===
 +  * CS data ^JM^
 +     * Anja's data    ## // PDT data that are currently being annotated for the extended coreference //
 +     * more 'user-friendly' texts    ## // texts that are currently in the LGame db//
 +   * **EN**
 +      * search the data that are available
 +=== Coding ===
 +  * utf-8
 +
 +=== Internal format ===
 +  * sgml ## //propose dtd file: include the element ''cor'' and its attribute ''src'' (manually, automatically, players) into csts.dtd; ve zkratce: u kazde slovni formy <f_id> bude povinne <l>, <t> a volitelne <cor src="man">, <cor src="Linh">, <cor src="player_id">, ..., <cor src="player_id">. Kdyz se bude s dokumentem hrat poprve, tak tam <cor src="player_id"> nebude zadne//
 +
 +=== (Pre)processing ===
 +  * tagging ## //see Tools needed below//
 +  * acr by Linh ## // dtto //
 +
 +=== Text handling ===
 +   * sentence by sentence
 +   * supervised selection of documents for a session 
 +
 +
 +
 +
 +
 +===== Scoring =====
 +   * ''pts_of_player_A = w1*(player_A's_output vs. manual_annotation(if available)) + w2*(player_A's_output vs. automatic_annotation) + w3*(player_A's_output vs. player_B's_output) + speed_pts''
 + 
 +// w1 by mela byt nejvyssi; w2 by mela urcite nejak zohlednit uspesnost automaticke procedury - uspesnost merenou na jakych datech?; w3: kdyz hracum budeme zobrazovat i ta slova, ktera oznacil protihrac, a ja je neoznacila, nebudeme je tim tlacit do vynucene shody? pro to, aby w3 bylo 'rozumne', bych zobrazovala pouze dvojice, na kterych se oba hraci shodli.//
 +
 +
 +
 +===== Output Data Needed =====
 +   * score list ## //player_id, pts, #sessions//
 +   * documents after the ''n''-th session consists of ''2*n'' players coreference annotation (some of them should be identical, the more identical the better)
 +   * session
 +      * player_A_id, player_B_id
 +      * document(s)
 +      * number of corrections by player_A and by player_B
 +      * corrections by player_A and by player_B
 +
 +
 +===== Design =====
 +   * What info to be displayed in the session?
 +      * session time = elapsed time + remaining time
 +      * how many sentences my partner has read so far 
 +      * running pts **???????**
 +   * Visualization of the coreference pairs
 +      * colors
 +      * arrows
 +      * ...
 +
 +
 +
 +
 +
 +===== Tools needed =====
 +   * tagger ^BH^ ## tool_chain (CAC2.0)
 +   * Linh's coreference resolution procedure ^PS^ ## What type of input data the Linh's procedure works with? ''tool_chain'' is going to be extended by the ''S'' option enabling to run Vasek Klimes' t-parser in a basic version, i.e. just t-tree and functors. 
 +  * conversion: csts <-> pml m_coref scheme
 +
 +

[ Back to the navigation ] [ Back to the content ]