Improve nsub and nn dependencies analysis

documentation/Cases_review.md

@@ -0,0 +1,81 @@
+
+Amélioration de nsubj/dobj avec instance_of
+===========================================
+
+#### instance_of + nsubj(pass)
+
+* nsubjpass + prep_in    : What language is spoken in Argentina? 
+* nsubj + dobj           : What actor married John F. Kennedy's sister?
+* nsubj + prep_by        : List movies directed by Spielberg
+* nsubjpass              : Which president has been killed by Oswald?
+* nsubjpass              : which book was authored by Victor Hugo
+
+#### instance_of + dobj
+
+* Which books did Suzanne Collins write?
+* How many films did Ingmar Bergman make?
+* How many children does Barack Obama have?
+* How many gas stations are there in the United States?
+
+#### nsubj avec verbe nécessaire
+
+* What is the most beautiful country in Europe?
+* Who was the first Taiwanese President?
+* What was the monetary value of the Nobel Peace Prize in 1989? 
+* When was Benjamin Disraeli prime minister?
+* nsubjpass : Where was Ulysses S. Grant born?
+* nsubjpass : Where is Inoco based?
+* What was the first Gilbert and Sullivan opera?
+* Where is the ENS of Lyon?
+* What did Bob write ?
+* Who is the author of Sea and Sky?
+* Is there a ghost in my house
+* Are there computers in your room
+
+#### Question word nsubj
+
+No subject after preprocessing
+
+* Who wrote the song, "Stardust"?
+* Who invented the hula hoop?
+* Who elected the president ?
+* Who was killed by Oswald?
+
+#### Autres
+
+* tmod : which day was the president born
+* prep_of : Of which country is Paris the capital? > mal parsé
+* prep_in : In which countries is the Lake Victoria? https://www.google.fr/webhp?sourceid=chrome-instant&ion=1&espv=2&ie=UTF-8#q=%22in+which+countries%22
+* prep_from : From which country is Alan Turing?
+
+_________________________________________________________________________________________________________________________________
+
+
+Exists
+======
+
+* Is there a ghost in my house
+* Is there a pilot in the plane
+* Is there a capital in France
+* Is there a king of england > https://www.wikidata.org/wiki/Q18810062
+* http://english.stackexchange.com/questions/34353/is-there-versus-are-there
+* Are there any articles available on the subject?
+* Are there computers in your room
+* Does a king of England exist?
+
+_________________________________________________________________________________________________________________________________
+
+Semi question words
+===================
+
+* Show me Star Wars movies > mal parsé
+* List movies directed by Spielberg
+* List books by Roald Dahl
+* List albums of Pink Floyd
+* List films with Jack Nicholson
+* List of US presidents
+* List of presidents of France
+* Give me the capital of France
+* Give the capital of France
+* Give us the capital of France
+* list of president of usa > mal parsé

documentation/General_questions.md

@@ -1,14 +1,12 @@
 General
 =======
 
-* Yes/no question: product dobj relation?
 * verb+ing: do sthg special (look POS tag)? : What did Richard Feynman say upon hearing he would receive the Nobel Prize in Physics?
 * If nounification becomes powerful enough: use it to analyse superlative (biggest > size...)
 * Multiple words : Where is Inoco based? > base + place = base place :( >> en fait "base" se nounifie en "place" ?
 * Article : enlever numéro apparaissant dans noeud (et idem pour arbre) / enlever encadrement
 * data model : autoriser des listes de prédicats dans les sort ?
 * réecrire demo3 pour le rendre dépendant de DependencyTree
-* t5 peut être enlevé ?
 * Dans le question word processing (et plus généralement) : les connecteurs ne sont pas uniquement les 1000. Les 1000 prennent les conj mais pas le superlatives.
 * Travailler sur la forme normalisée? 
     >> garantir qu'en entrée de normalize chaque noeud contient une seule alternative
@@ -20,6 +18,19 @@ General
 * Multiple predicates pour les sort
 * Tell me where the DuPont company is located. Name the Ranger who was always after Yogi Bear.
 * How do you solve "Rubik's Cube"? > en quoi est transformé how
+* réduire le nb de map, ajouter + d'infos
+* autres auxiliaire (have) : What dictator has the nickname "El Maximo"
+* propagation de types : nsubjRule + qw in strongQuestionWord = R5s
+* Who was the leader of the Branch Davidian Cult confronted by the FBI in Waco, Texas in 1993?    >> gros sujet
+* Where is Inoco based?                                                                           >> revoir la nounification associée
+* Who Clinton defeated?                                                                           >> prq nounification échoue ? non lemmatizé ?
+* Rapprocher/renommer les règles R.. similaires
+* __How many__ : opérateur de comptage 
+   > How many films did Ingmar Bergman make?
+   > How many children does Barack Obama have?
+   > How many gas stations are there in the United States?
+   > cf instance_of sur dobj >> on récupère la liste produite en sortie et on renvoie sa taille
+   > How much did Mercury spend on advertising in 1993?
 
 Remarks
 =======
@@ -33,7 +44,7 @@ Gestion des prep(c)_x
 =====================
 
 go -prep_to-> ... = go to -prep->
-What is Frozen based on?
+* What is Frozen based on?
 
 * What two US biochemists won the Nobel Prize in medicine in 1992?
 
@@ -49,55 +60,178 @@ Superlative
 Yes/No questions
 ================
 
-Exists
-======
+* Yes/no question: product dobj relation?
+* nsubj + prep_from                 : Are you from Germany?                     > (you,origin,Germany) > yes/no : (subj | pred:be from, do..live | cpt)
 
-* Is there a ghost in my house
-* Is there a pilot in the plane
-* Is there a capital in France
-* Is there a king of england > https://www.wikidata.org/wiki/Q18810062
-* http://english.stackexchange.com/questions/34353/is-there-versus-are-there
-* Are there any articles available on the subject?
-* Are there computers in your room
-* Does a king of England exist?
+Conjonction
+===========
 
-Semi question words
+Mauvais : 
+  * What was the first Gilbert and Sullivan opera?
+
+Exemples :
+----------
+* Who makes and distributes bells?
+* Who is the author of Sea and Sky?
+* What percentage of the world's plant and animal species can be found in the Amazon forests?
+* Good: Who is section manager for guidance and control systems at JPL?
+* Bad: How many people did the United Nations commit to help restore order and distribute humanitarian relief in Somalia in September 1992?
+* Bad: Which Italian city is home to the Cathedral of Santa Maria del Fiore or the Duomo?
+
+Problem with merging:
+---------------------
+* What is the length of border between the Ukraine and Russia?
+
+Comment construire les sous arbres
+----------------------------------
+* What was the first Gilbert and Sullivan opera?
+* When was General Manuel Noriega ousted as the leader of Panama and turned over to U.S. authorities?
+* When did Princess Diana and Prince Charles get married?
+* When did the royal wedding of Prince Andrew and Fergie take place?
+* ++ How many people did the United Nations commit to help restore order and distribute humanitarian relief in Somalia in September 1992?
+    >> peut être propager les prep après ?
+    >> même problème que pour les nn
+
+Merge nn with the 2 nodes if nn above them:
+ - When did Princess Diana and Charles get married?
+ - When did Princess Diana and Prince Charles get married?
+ - Who is section manager for guidance and control systems at JPL?
+
+_________________________________________________________________________________________________________________________________
+_________________________________________________________________________________________________________________________________
+
+Améliorer la MWE recognition
+============================
+
+Rattraper un mauvais parsing:
+  * who is the president of the United states of america
+  * Where is the ENS of Lyon? (merge car majuscule?)
+
+Good:
+  * Who is the United States president
+  * What was the first Gilbert and Sullivan opera?
+  * Obama is the United States president.
+
+Amod:
+  * Who is the French president? >> nécessite avant de transformer French en France
+  * Who was the first Taiwanese President?
+
+What organization was founded by the Rev. Jerry Falwell? >> tagger Rev car majuscule
+
+_________________________________________________________________________________________________________________________________
+
+Traitement des prep
 ===================
 
-* Show me Star Wars movies
+Passer prep en Rnew
+  * verbe auxiliaire : 
+   - 
+  * verbe non auxiliaire : 
+   - List movies directed by Spielberg
+   - What language is spoken in Argentina? :(
+   - What kings ruled on France?
+   - Who was born on 1984?
+  * nom : 
+   - List of books by Roald Dahl
+   - president of France
+
+_________________________________________________________________________________________________________________________________
+
+### nsubj
+
+R5
+==
+
+* Where does the president live?
+
+R3
+==
+
+R5 ou R3
+========
+
+* What did George Orwell write?
+* Which books did Suzanne Collins write?
+
+### nsubpass
+
+R5
+==
+
+R3
+==
+
+R5 ou R3
+========
+
+* Where was Ulysses S. Grant born?
+* Where is Inoco based?
+
+### agent
+
+R5
+==
+
+R3
+==
+
+R5 ou R3
+========
+
+* Who was killed by Oswald?
+* Which president has been killed by Oswald?
+* Which books were authored by Victor Hugo?
+
+----------------
+
+### dobj
+
+R5
+==
+
+R3
+==
+
+R5 ou R3
+========
+
+* Who developed Microsoft?
+* What actor married John F. Kennedy's sister?
+* Who has written "The Hitchhiker's Guide to the Galaxy"?
+* Who wrote the song, "Stardust"?
+* Who invented the hula hoop?
+* Who elected the president ?
+* Who killed Gandhi?
+
+### prep (+ V)
+
+R5
+==
+
+R3
+==
+
+* Which kings ruled on France
 * List movies directed by Spielberg
-* List books by Roald Dahl
-* List albums of Pink Floyd
-* List films with Jack Nicholson
-* List of US presidents
-* List of presidents of France
-* Give me the capital of France
-* Give the capital of France
-* Give us the capital of France
-
-Racine à fils multiples
-=======================
 
-* nsubj + prep_from                 : Are you from Germany?                     > (you,origin,Germany) > yes/no : (subj | pred:be from, do..live | cpt)
-* nsubj + prep_by                   : List movies directed by Spielberg
-* prep_of + prep_of                 : list of president of usa
-* nsubj + prep_by                   : List movies directed by Spielberg
+R5 ou R3
+========
+
+* What language is spoken in Argentina?
+* Who followed Willy Brandt as chancellor of the Federal Republic of Germany?
+* Who was born on 1984
+
+----------------
 
+The animal | lives in | the farm.
+ Subject     Predicate   Object    >> ( animal , residence , farm )
 
-instance of:
+The animal | lives in | the farm.
+  Object     Predicate   Subject   >> ( farm , inhabitant , animal )
 
-* prep_from + prep_to + prep_on     : carpool from Lyon to Paris on December 31 > (?, instance of, carpool) ∩ (?, from, Paris) ∩ (?, to, Lyon) ∩ (?,day, December 31st)
-* nsubjpass + prep_in               : What language is spoken in Argentina?     > (Argentina, language, ?)
-* nsubj + dobj                      : Which books did Suzanne Collins write?    > (Suzanne Collins, author, ?) + typage "book" sur ?
-* nsubj + dobj (+ do)               : What albums did Pearl Jam record?
-* nsubj + dobj                      : What dictator has the nickname "El Maximo"?
-* nsubj + dobj                      : What actor married John F. Kennedy's sister? > (?, instance of, actor) ∩ (?, wife, (John F. Kennedy, sister, ?))
-* nsubj + prep_in                   : How many gas stations are there in the United States?
+---------------
 
-* voir Problematic questions dans hierarchy review
+processQuestionInfo dans questionWordProcessing doit être le seul habilité à affaiblir une règle en R2 (ou R3 bis) (mais pas en R0: where is the residence)
+dependency analysis pose un R5/R3 puis processQuestionInfo affaiblie les dépendances de plus haut niveau s'il trouve l'info en-dessous
 
-Amélioration des question maps
-==============================
 
-* How much : ajouter cost
-* Plus généralement : réduire le nb de map, ajouter + d'infos