Merge branch 'main' into bayes_rule

Author: x-tabdeveloping

README.md

@@ -34,20 +34,17 @@ model.hierarchy.divide_children(n_subtopics=3)
 print(model.hierarchy)
 ```
 
-<div style="background-color: #F5F5F5; padding: 10px; padding-left: 20px; padding-right: 20px;">
-<tt style="font-size: 11pt">
-<b>Root </b><br>
-├── <b style="color: blue">0</b>: windows, dos, os, disk, card, drivers, file, pc, files, microsoft <br>
-│   ├── <b style="color: magenta">0.0</b>: dos, file, disk, files, program, windows, disks, shareware, norton, memory <br>
-│   ├── <b style="color: magenta">0.1</b>: os, unix, windows, microsoft, apps, nt, ibm, ms, os2, platform <br>
-│   └── <b style="color: magenta">0.2</b>: card, drivers, monitor, driver, vga, ram, motherboard, cards, graphics, ati <br>
-└── <b style="color: blue">1</b>: atheism, atheist, atheists, religion, christians, religious, belief, christian, god, beliefs <br>
-.    ├── <b style="color: magenta">1.0</b>: atheism, alt, newsgroup, reading, faq, islam, questions, read, newsgroups, readers <br>
-.    ├── <b style="color: magenta">1.1</b>: atheists, atheist, belief, theists, beliefs, religious, religion, agnostic, gods, religions <br>
-.    └── <b style="color: magenta">1.2</b>: morality, bible, christian, christians, moral, christianity, biblical, immoral, god, religion <br>
-</tt>
-</div>
-
+```
+Root
+├── windows, dos, os, disk, card, drivers, file, pc, files, microsoft
+│   ├── 0.0: dos, file, disk, files, program, windows, disks, shareware, norton, memory
+│   ├── 0.1: os, unix, windows, microsoft, apps, nt, ibm, ms, os2, platform
+│   └── 0.2: card, drivers, monitor, driver, vga, ram, motherboard, cards, graphics, ati
+└── 1: atheism, atheist, atheists, religion, christians, religious, belief, christian, god, beliefs
+.    ├── 1.0: atheism, alt, newsgroup, reading, faq, islam, questions, read, newsgroups, readers
+.    ├── 1.1: atheists, atheist, belief, theists, beliefs, religious, religion, agnostic, gods, religions
+.    └── 1.2: morality, bible, christian, christians, moral, christianity, biblical, immoral, god, religion
+```
 
 #### FASTopic *(Experimental)*
 

pyproject.toml

@@ -6,7 +6,7 @@ line-length=79
 
 [tool.poetry]
 name = "turftopic"
-version = "0.5.1"
+version = "0.5.4"
 description = "Topic modeling with contextual representations from sentence transformers."
 authors = ["Márton Kardos <power.up1163@gmail.com>"]
 license = "MIT"

turftopic/models/decomp.py

@@ -6,7 +6,7 @@
 from sklearn.base import TransformerMixin
 from sklearn.decomposition import FastICA
 from sklearn.feature_extraction.text import CountVectorizer
-from sklearn.metrics.pairwise import euclidean_distances
+from sklearn.metrics.pairwise import cosine_similarity, euclidean_distances
 
 from turftopic.base import ContextualModel, Encoder
 from turftopic.vectorizer import default_vectorizer
@@ -42,6 +42,9 @@ class SemanticSignalSeparation(ContextualModel):
         If not specified, FastICA is used.
     max_iter: int, default 200
         Maximum number of iterations for ICA.
+    feature_importance: "axial", "angular" or "combined", default "combined"
+        Defines whether the word's position on an axis ('axial'), it's angle to the axis ('angular')
+        or their combination ('combined') should determine the word's importance for a topic.
     random_state: int, default None
         Random state to use so that results are exactly reproducible.
     """
@@ -55,10 +58,14 @@ def __init__(
         vectorizer: Optional[CountVectorizer] = None,
         decomposition: Optional[TransformerMixin] = None,
         max_iter: int = 200,
+        feature_importance: Literal[
+            "axial", "angular", "combined"
+        ] = "combined",
         random_state: Optional[int] = None,
     ):
         self.n_components = n_components
         self.encoder = encoder
+        self.feature_importance = feature_importance
         if isinstance(encoder, str):
             self.encoder_ = SentenceTransformer(encoder)
         else:
@@ -76,6 +83,20 @@ def __init__(
         else:
             self.decomposition = decomposition
 
+    def estimate_components(
+        self, feature_importance: Literal["axial", "angular", "combined"]
+    ) -> np.ndarray:
+        """Reestimates components based on the chosen feature_importance method."""
+        if feature_importance == "axial":
+            self.components_ = self.axial_components_
+        elif feature_importance == "angular":
+            self.components_ = self.angular_components_
+        elif feature_importance == "combined":
+            self.components_ = (
+                np.square(self.axial_components_) * self.angular_components_
+            )
+        return self.components_
+
     def fit_transform(
         self, raw_documents, y=None, embeddings: Optional[np.ndarray] = None
     ) -> np.ndarray:
@@ -96,10 +117,30 @@ def fit_transform(
             console.log("Vocabulary encoded.")
             status.update("Estimating term importances")
             vocab_topic = self.decomposition.transform(vocab_embeddings)
-            self.components_ = vocab_topic.T
+            self.axial_components_ = vocab_topic.T
+            if self.feature_importance == "axial":
+                self.components_ = self.axial_components_
+            elif self.feature_importance == "angular":
+                self.components_ = self.angular_components_
+            elif self.feature_importance == "combined":
+                self.components_ = (
+                    np.square(self.axial_components_)
+                    * self.angular_components_
+                )
             console.log("Model fitting done.")
         return doc_topic
 
+    @property
+    def angular_components_(self):
+        """Reweights words based on their angle in ICA-space to the axis
+        base vectors.
+        """
+        word_vectors = self.axial_components_.T
+        n_topics = self.axial_components_.shape[0]
+        axis_vectors = np.eye(n_topics)
+        cosine_components = cosine_similarity(axis_vectors, word_vectors)
+        return cosine_components
+
     def transform(
         self, raw_documents, embeddings: Optional[np.ndarray] = None
     ) -> np.ndarray:
@@ -211,8 +252,8 @@ def concept_compass(
                 raise ValueError(
                     f"{topic_y} is not a valid topic name or index."
                 ) from e
-        x = self.components_[topic_x]
-        y = self.components_[topic_y]
+        x = self.axial_components_[topic_x]
+        y = self.axial_components_[topic_y]
         vocab = self.get_vocab()
         points = np.array(list(zip(x, y)))
         xx, yy = np.meshgrid(