Added functionality to merge custom words list using full verbs

orangecontrib/storynavigation/modules/tagging.py

@@ -19,14 +19,16 @@ class Tagger:
 
     Args:
         n_segments (int): Number of segments to split each story into.
+        use_infinitives (bool): Whether to use infinitives for verbs.
     """
-    def __init__(self, lang, n_segments, remove_stopwords, text_tuples, custom_tags_and_word_column=None, callback=None):
+    def __init__(self, lang, n_segments, remove_stopwords, text_tuples, custom_tags_and_word_column=None, callback=None,use_infinitives=False):
         self.text_tuples = text_tuples
         self.lang = lang
         self.n_segments = n_segments
         self.remove_stopwords = remove_stopwords
         self.custom_tags = None
         self.word_column = None
+        self.use_infinitives = use_infinitives
         # any new column name added below should also be added to variable TAGGING_DATAFRAME_COLUMNNAMES in constants.py
         self.complete_data_columns = ['storyid', 'sentence', 'token_text', 'token_start_idx', 'token_end_idx', 'story_navigator_tag', 'spacy_tag', 'spacy_finegrained_tag', 'spacy_dependency', 'spacy_ne', 'spacy_lemma', 'spacy_head_text', 'spacy_head_idx', 'is_pronoun_boolean', 'is_sentence_subject_boolean', 'active_voice_subject_boolean', 'associated_action']
 
@@ -76,18 +78,30 @@ def __process_stories(self, nlp, text_tuples, callback):
             if callback:
                 callback((c / len(text_tuples) * 100))
 
+        # Process custom tags and word column if provided
         if self.custom_tags is not None and self.word_column is not None:
             collection_df['custom_' + self.word_column] = collection_df['token_text'].str.lower()
             collection_df['custom_' + self.word_column] = collection_df['custom_' + self.word_column].str.lstrip('0123456789@#$!“"-')
 
             if self.use_infinitives:
-                collection_df = pd.merge(collection_df, self.custom_tags, left_on='custom_' + self.word_column, right_on=self.word_column, how='left')
-                collection_df = collection_df.drop(columns=[self.word_column])
+                collection_df['custom_' + self.word_column] = collection_df['spacy_lemma'].str.lower()               
+
+            else:
+                # If not using infinitives, simply copy the custom word column
+                collection_df['custom_' + self.word_column] = collection_df['custom_' + self.word_column]
+
+            # Merge the custom tags
+            collection_df = pd.merge(collection_df, self.custom_tags, left_on='custom_' + self.word_column, right_on=self.word_column, how='left')
+            collection_df = collection_df.drop(columns=[self.word_column])
+
         else:
             collection_df['token_text_lowercase'] = collection_df['token_text'].str.lower()
 
+        # Clean up associated action columns
         collection_df['associated_action'] = collection_df['associated_action'].str.lstrip('0123456789@#$!“"-')
         collection_df['associated_action_lowercase'] = collection_df['associated_action'].str.lower()
+
+        # Add language column and word count
         lang_col_values = [self.lang] * len(collection_df)
         collection_df['lang'] = lang_col_values
         story_wordcount_values = self.__calculate_story_wordcounts(collection_df)

orangecontrib/storynavigation/widgets/OWSNTagger.py

@@ -37,13 +37,15 @@ class Error(OWWidget.Error):
     word_column = 'word'
     n_segments = 1
     remove_stopwords = constants.YES
+    use_infinitives = Setting(False)
 
     def __init__(self):
         super().__init__()
         ConcurrentWidgetMixin.__init__(self)
         self.stories = None # initialise list of documents (corpus)
         self.custom_tag_dict = None
         self.custom_tag_dict_columns = ['']
+        self.use_infinitives = False
 
         size_policy = QSizePolicy(QSizePolicy.Maximum, QSizePolicy.Maximum)
         self.controlArea.setSizePolicy(size_policy)
@@ -112,6 +114,17 @@ def __init__(self):
                 }
             """
         )
+
+        self.infinitives_checkbox = gui.checkBox(
+            widget=self.controlArea,
+            master=self,
+            value='use_infinitives',
+            label='Use infinitives to merge custom words',
+            callback=self.on_infinitives_changed,
+            sizePolicy=QSizePolicy(QSizePolicy.Maximum, QSizePolicy.Maximum)
+        )
+
+        self.controlArea.layout().addWidget(self.infinitives_checkbox)
 
     @Inputs.stories
     def set_stories(self, stories=None):
@@ -163,6 +176,10 @@ def reset_widget(self):
     def on_done(self, result) -> None:
         self.Outputs.dataset_level_data.send(table_from_frame(self.tagger.complete_data))                
 
+    def on_infinitives_changed(self):
+        #add any additional logic here if needed
+        pass
+
     def run(self, lang, n_segments, remove_stopwords, text_tuples, tuple, state: TaskState):
         def advance(progress):
             if state.is_interruption_requested():
@@ -171,7 +188,8 @@ def advance(progress):
 
         self.tagger = Tagger(
             lang=lang, n_segments=n_segments, remove_stopwords=remove_stopwords, text_tuples=text_tuples, 
-            custom_tags_and_word_column=tuple, callback=advance)
+            custom_tags_and_word_column=tuple, callback=advance,
+            use_infinitives=self.use_infinitives)
 
         return self.tagger.complete_data