From 88ed052e4409920e3c16ca6e75c255da49e96211 Mon Sep 17 00:00:00 2001 From: Jake Lever Date: Fri, 3 Mar 2023 20:57:43 +0000 Subject: [PATCH] More updated to vectorizer test data --- tests/data/vectorizer/expected.json | 66 +++++++++++++---------------- 1 file changed, 30 insertions(+), 36 deletions(-) diff --git a/tests/data/vectorizer/expected.json b/tests/data/vectorizer/expected.json index 75faac7..b3f54d1 100644 --- a/tests/data/vectorizer/expected.json +++ b/tests/data/vectorizer/expected.json @@ -173,8 +173,8 @@ "bigrams_vgypkemhjr_.": 0.81649658, "bigrams_with_vgypkemhjr": 0.81649658, "bigrams_zgwivlcmly_.": 0.78013025, + "dependencypathelements_amod": 2.0, "dependencypathelements_attr": 4.0, - "dependencypathelements_dobj": 2.0, "dependencypathelements_nsubj": 6.0, "dependencypathelements_nsubjpass": 2.0, "dependencypathelements_pobj": 8.0, @@ -234,8 +234,8 @@ "bigrams_vgypkemhjr_.": 0.81649658, "bigrams_with_vgypkemhjr": 0.81649658, "bigrams_zgwivlcmly_.": 0.0, + "dependencypathelements_amod": 4.0, "dependencypathelements_attr": 4.0, - "dependencypathelements_dobj": 4.0, "dependencypathelements_nsubj": 8.0, "dependencypathelements_nsubjpass": 2.0, "dependencypathelements_pobj": 10.0, @@ -289,44 +289,41 @@ "bigrams_zgwivlcmly_can": 2.28320249, "bigrams_zkrkzlyfef_inhibition": 2.28320249, "dependencypathelements_0_1_acl": 8.0, - "dependencypathelements_0_1_advmod": 4.0, "dependencypathelements_0_1_agent": 8.0, "dependencypathelements_0_1_compound": 8.0, "dependencypathelements_0_1_conj": 4.0, - "dependencypathelements_0_1_dobj": 16.0, + "dependencypathelements_0_1_dobj": 12.0, "dependencypathelements_0_1_nsubj": 4.0, - "dependencypathelements_0_1_nsubjpass": 4.0, + "dependencypathelements_0_1_nsubjpass": 8.0, "dependencypathelements_0_1_pobj": 8.0, "dependencypathelements_0_2_acl": 8.0, - "dependencypathelements_0_2_advmod": 4.0, "dependencypathelements_0_2_agent": 8.0, "dependencypathelements_0_2_compound": 8.0, "dependencypathelements_0_2_conj": 4.0, - "dependencypathelements_0_2_dobj": 16.0, + "dependencypathelements_0_2_dobj": 12.0, "dependencypathelements_0_2_nsubj": 4.0, - "dependencypathelements_0_2_nsubjpass": 4.0, + "dependencypathelements_0_2_nsubjpass": 8.0, "dependencypathelements_0_2_pobj": 8.0, "dependencypathelements_1_2_acl": 8.0, - "dependencypathelements_1_2_advmod": 4.0, "dependencypathelements_1_2_agent": 8.0, "dependencypathelements_1_2_compound": 8.0, "dependencypathelements_1_2_conj": 4.0, - "dependencypathelements_1_2_dobj": 16.0, + "dependencypathelements_1_2_dobj": 12.0, "dependencypathelements_1_2_nsubj": 4.0, - "dependencypathelements_1_2_nsubjpass": 4.0, + "dependencypathelements_1_2_nsubjpass": 8.0, "dependencypathelements_1_2_pobj": 8.0, - "dependencypathnearselectedtoken_0_advmod": 2.0, "dependencypathnearselectedtoken_0_compound": 4.0, + "dependencypathnearselectedtoken_0_conj": 2.0, "dependencypathnearselectedtoken_0_nsubj": 2.0, - "dependencypathnearselectedtoken_0_nsubjpass": 2.0, - "dependencypathnearselectedtoken_1_advmod": 2.0, + "dependencypathnearselectedtoken_0_nsubjpass": 4.0, "dependencypathnearselectedtoken_1_compound": 4.0, + "dependencypathnearselectedtoken_1_conj": 2.0, "dependencypathnearselectedtoken_1_nsubj": 2.0, - "dependencypathnearselectedtoken_1_nsubjpass": 2.0, - "dependencypathnearselectedtoken_2_advmod": 2.0, + "dependencypathnearselectedtoken_1_nsubjpass": 4.0, "dependencypathnearselectedtoken_2_compound": 4.0, + "dependencypathnearselectedtoken_2_conj": 2.0, "dependencypathnearselectedtoken_2_nsubj": 2.0, - "dependencypathnearselectedtoken_2_nsubjpass": 2.0, + "dependencypathnearselectedtoken_2_nsubjpass": 4.0, "ngrams_betweenentities_0_1_and": 1.46201744, "ngrams_betweenentities_0_1_be": 2.08991136, "ngrams_betweenentities_0_1_by": 2.08991136, @@ -395,44 +392,41 @@ "bigrams_zgwivlcmly_can": 2.28320249, "bigrams_zkrkzlyfef_inhibition": 2.28320249, "dependencypathelements_0_1_acl": 8.0, - "dependencypathelements_0_1_advmod": 4.0, "dependencypathelements_0_1_agent": 8.0, "dependencypathelements_0_1_compound": 8.0, "dependencypathelements_0_1_conj": 4.0, - "dependencypathelements_0_1_dobj": 16.0, + "dependencypathelements_0_1_dobj": 12.0, "dependencypathelements_0_1_nsubj": 4.0, - "dependencypathelements_0_1_nsubjpass": 4.0, + "dependencypathelements_0_1_nsubjpass": 8.0, "dependencypathelements_0_1_pobj": 8.0, "dependencypathelements_0_2_acl": 8.0, - "dependencypathelements_0_2_advmod": 4.0, "dependencypathelements_0_2_agent": 8.0, "dependencypathelements_0_2_compound": 8.0, "dependencypathelements_0_2_conj": 4.0, - "dependencypathelements_0_2_dobj": 16.0, + "dependencypathelements_0_2_dobj": 12.0, "dependencypathelements_0_2_nsubj": 4.0, - "dependencypathelements_0_2_nsubjpass": 4.0, + "dependencypathelements_0_2_nsubjpass": 8.0, "dependencypathelements_0_2_pobj": 8.0, "dependencypathelements_1_2_acl": 8.0, - "dependencypathelements_1_2_advmod": 4.0, "dependencypathelements_1_2_agent": 8.0, "dependencypathelements_1_2_compound": 8.0, "dependencypathelements_1_2_conj": 4.0, - "dependencypathelements_1_2_dobj": 16.0, + "dependencypathelements_1_2_dobj": 12.0, "dependencypathelements_1_2_nsubj": 4.0, - "dependencypathelements_1_2_nsubjpass": 4.0, + "dependencypathelements_1_2_nsubjpass": 8.0, "dependencypathelements_1_2_pobj": 8.0, - "dependencypathnearselectedtoken_0_advmod": 2.0, "dependencypathnearselectedtoken_0_compound": 4.0, + "dependencypathnearselectedtoken_0_conj": 2.0, "dependencypathnearselectedtoken_0_nsubj": 2.0, - "dependencypathnearselectedtoken_0_nsubjpass": 2.0, - "dependencypathnearselectedtoken_1_advmod": 2.0, + "dependencypathnearselectedtoken_0_nsubjpass": 4.0, "dependencypathnearselectedtoken_1_compound": 4.0, + "dependencypathnearselectedtoken_1_conj": 2.0, "dependencypathnearselectedtoken_1_nsubj": 2.0, - "dependencypathnearselectedtoken_1_nsubjpass": 2.0, - "dependencypathnearselectedtoken_2_advmod": 2.0, + "dependencypathnearselectedtoken_1_nsubjpass": 4.0, "dependencypathnearselectedtoken_2_compound": 4.0, + "dependencypathnearselectedtoken_2_conj": 2.0, "dependencypathnearselectedtoken_2_nsubj": 2.0, - "dependencypathnearselectedtoken_2_nsubjpass": 2.0, + "dependencypathnearselectedtoken_2_nsubjpass": 4.0, "ngrams_betweenentities_0_1_and": 1.46201744, "ngrams_betweenentities_0_1_be": 2.08991136, "ngrams_betweenentities_0_1_by": 2.08991136, @@ -504,32 +498,32 @@ "dependencypathnearselectedtoken_1_nsubjpass": 1.0 }, "test_vectorizer_dependencyPathEdges_1": { + "dependencypathelements_amod": 2.0, "dependencypathelements_attr": 4.0, - "dependencypathelements_dobj": 2.0, "dependencypathelements_nsubj": 6.0, "dependencypathelements_nsubjpass": 2.0, "dependencypathelements_pobj": 8.0, "dependencypathelements_prep": 8.0 }, "test_vectorizer_dependencyPathEdges_2": { + "dependencypathelements_amod": 2.0, "dependencypathelements_attr": 4.0, - "dependencypathelements_dobj": 2.0, "dependencypathelements_nsubj": 6.0, "dependencypathelements_nsubjpass": 2.0, "dependencypathelements_pobj": 8.0, "dependencypathelements_prep": 8.0 }, "test_vectorizer_dependencyPathEdges_noTFIDF_1": { + "dependencypathelements_amod": 2.0, "dependencypathelements_attr": 4.0, - "dependencypathelements_dobj": 2.0, "dependencypathelements_nsubj": 6.0, "dependencypathelements_nsubjpass": 2.0, "dependencypathelements_pobj": 8.0, "dependencypathelements_prep": 8.0 }, "test_vectorizer_dependencyPathEdges_noTFIDF_2": { + "dependencypathelements_amod": 2.0, "dependencypathelements_attr": 4.0, - "dependencypathelements_dobj": 2.0, "dependencypathelements_nsubj": 6.0, "dependencypathelements_nsubjpass": 2.0, "dependencypathelements_pobj": 8.0,