nouveau fichier : init/sql/mainList.sql

Raw SQL to create lists.

	nouveau fichier : init/sql/cooccurrences.sql
Raw SQL to create cooccurrence from lists.

All theses SQL need to be checked before production.

init/sql/cooccurrences.sql

+---- BASIQUE calcul des cooccurrences en ne tenant pas compte des stems équivalents
+--
+--	SELECT
+--	-- %d as node_id,
+--	ngX.id,
+--	ngY.id,
+--	COUNT(*) AS score
+--FROM
+--	node_node AS n  -- the nodes who are direct children of the corpus
+--
+--INNER JOIN
+--	node_node_ngram AS nngX ON nngX.node_id = n.id  --  list of ngrams contained in the node
+--INNER JOIN
+--	node_node_ngram AS mainlistX ON mainlistX.ngram_id = nngX.ngram_id -- list of ngrams contained in the mainlist and in the node
+--INNER JOIN
+--	node_ngram AS ngX ON ngX.id = mainlistX.ngram_id -- ngrams which are in both
+--
+--INNER JOIN
+--	node_node_ngram AS nngY ON nngY.node_id = n.id
+--INNER JOIN
+--	node_node_ngram AS mainlistY ON mainlistY.ngram_id = nngY.ngram_id
+--INNER JOIN
+--	node_ngram AS ngY ON ngY.id = mainlistY.ngram_id
+--
+--WHERE
+--	n.parent_id = 1298
+--AND
+--	n.type_id = 5
+--AND
+--	mainlistX.node_id = 1382
+--AND
+--	mainlistY.node_id = 1382
+--AND
+--	nngX.ngram_id < nngY.ngram_id   --  so we only get distinct pairs of ngrams
+--
+--GROUP BY
+--	ngX.id,
+--	ngX.terms,
+--	ngY.id,
+--	ngY.terms
+--
+--ORDER BY score DESC
+--LIMIT 3
+--;
+--
+
+-- calcul des cooccurrences en tenant compte des stems équivalents
+
+	SELECT
+	-- %d as node_id,
+	ngX.id,
+	ngY.id,
+	COUNT(*) AS score
+FROM
+	node_node AS n  -- the nodes who are direct children of the corpus
+
+INNER JOIN
+	node_node_ngram AS nngX ON nngX.node_id = n.id  --  list of ngrams contained in the node
+INNER JOIN
+	node_node_ngram AS mainlistX ON mainlistX.ngram_id = nngX.ngram_id -- list of ngrams contained in the mainlist and in the node
+INNER JOIN
+	node_ngram AS ngX ON ngX.id = mainlistX.ngram_id -- ngrams which are in both
+LEFT JOIN
+	node_nodengramngram AS nggXX ON nggXX.node_id = 94
+	AND nggXX.ngramx_id = ngX.id
+LEFT JOIN
+	node_nodengramngram AS nggXY ON nggXY.node_id = 94
+	AND nggXY.ngramy_id = nggXY.ngramy_id
+	AND nggXY.ngramx_id < nggXY.ngramx_id
+
+INNER JOIN
+	node_node_ngram AS nngY ON nngY.node_id = n.id
+INNER JOIN
+	node_node_ngram AS mainlistY ON mainlistY.ngram_id = nngY.ngram_id
+INNER JOIN
+	node_ngram AS ngY ON ngY.id = mainlistY.ngram_id
+LEFT JOIN
+	node_nodengramngram AS nggYX ON nggYX.node_id = 94
+	AND nggYX.ngramx_id = ngY.id
+LEFT JOIN
+	node_nodengramngram AS nggYY ON nggYY.node_id = 94
+	AND nggYX.ngramy_id = nggYY.ngramy_id
+	AND nggYX.ngramx_id < nggYY.ngramx_id
+
+WHERE
+	n.parent_id = 1298
+AND
+	n.type_id = 5
+AND
+	mainlistX.node_id = 1382
+AND
+	mainlistY.node_id = 1382
+AND
+	nngX.ngram_id < nngY.ngram_id   --  so we only get distinct pairs of ngrams
+--AND
+--	nggYY.id is NULL
+--AND
+--	nggXY.id is NULL
+	
+GROUP BY
+	ngX.id,
+	ngX.terms,
+	ngY.id,
+	ngY.terms
+
+ORDER BY score DESC
+LIMIT 3
+;
+

init/sql/mainList.sql

+
+-- select tous les ngrams distincts de la miam list
+SELECT count(*) FROM
+(
+SELECT ngram_id FROM node_node_ngram 
+WHERE node_id = 1380  --> node.id de la miam list
+GROUP BY ngram_id
+) as global
+	;
+
+-- select tous les ngrams d'un corpus ayant un stem
+SELECT count(*) FROM
+(
+SELECT ngramx_id FROM node_nodengramngram as ng
+
+INNER JOIN node_node_ngram as nn
+ON nn.ngram_id = ng.ngramx_id
+
+INNER JOIN node_node as n
+ON n.id = nn.node_id
+AND n.parent_id = 1298 --> node.id du corpus 
+
+WHERE ng.node_id = 94  --> node.id de la stem list
+GROUP BY ng.ngramx_id
+) as global
+	;
+
+
+
+--- select uniquement tous les ngrams distincts qui ont des stems équivalents
+-- LEFT JOIN inclusif des ngrams qui on un stem
+-- LEFT JOIN exclusif des ngrams qui on un stem en commun
+
+select count(*) from
+(
+SELECT ngram_id FROM node_node_ngram  as nn
+
+INNER JOIN node_node as n
+ON nn.node_id = n.id
+AND n.parent_id = 1298 --> node.id du corpus 
+
+LEFT JOIN node_nodengramngram AS nx
+ON nx.node_id = 94 --> node.id Stem
+AND nx.ngramx_id = nn.ngram_id 
+
+LEFT JOIN node_nodengramngram AS ny
+ON nx.ngramy_id = ny.ngramy_id 
+AND nx.node_id = 94 --> node.id Stem
+AND nx.ngramx_id < ny.ngramx_id  --> pour supprimer les doublons
+
+WHERE nn.node_id = 1380 --> node.id de la miam list
+-- AND ny.id is NULL
+
+GROUP BY nn.ngram_id, nx.ngramx_id --, ny.ngramx_id
+) as global
+;