Updated README.md + changed default spaCy model to "Large".

Author: Halvani

Demo_POSNoise.ipynb

@@ -2,48 +2,16 @@
  "cells": [
   {
    "cell_type": "markdown",
-   "id": "1a592652",
+   "id": "50297305",
    "metadata": {},
    "source": [
     "## POSNoise"
    ]
   },
   {
    "cell_type": "code",
-   "execution_count": 1,
-   "id": "f3b4b9ca",
-   "metadata": {},
-   "outputs": [
-    {
-     "name": "stdout",
-     "output_type": "stream",
-     "text": [
-      "[POSNoise] Loading spaCy model 'en_core_web_sm' (disabled: parser, ner)...\n",
-      "[POSNoise] Loaded 'en_core_web_sm'.\n",
-      "The original dataset contains two partitions comprising sockpuppets and non-sockpuppets cases.\n",
-      "The @ # Ø two # Ø # and @@@ #.\n"
-     ]
-    }
-   ],
-   "source": [
-    "import posnoise\n",
-    "from posnoise import POSNoise"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 2,
-   "id": "9555ed7f",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "posnoise_instance = POSNoise(spacy_model_size=posnoise.core.SpacyModelSize.Large)"
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 4,
-   "id": "48f4addb",
+   "execution_count": 5,
+   "id": "f7dc478d",
    "metadata": {
     "scrolled": true
    },
@@ -52,18 +20,22 @@
      "name": "stdout",
      "output_type": "stream",
      "text": [
-      "I love python !\n",
-      "I Ø # !\n"
+      "Fitzgerald made her first tour of Australia in July 1954 for the Australian-based American promoter Lee Gordon.\n",
+      "§ made her first # of § in § µ for the §-Ø @ # § §.\n"
      ]
     }
    ],
    "source": [
     "import posnoise\n",
     "from posnoise import POSNoise\n",
     "\n",
-    "posnoise_instance = POSNoise(spacy_model_size=posnoise.core.SpacyModelSize.Large)\n",
+    "# By default POSNoise loads the \"Large\" spaCy model\n",
+    "posnoise_instance = POSNoise()\n",
+    "    \n",
+    "# In case you want specify another model, just set it accordingly e.g. \n",
+    "posnoise_instance = POSNoise(spacy_model_size=posnoise.core.SpacyModelSize.Medium)\n",
     "\n",
-    "document = \"I love python !\"\n",
+    "document = \"Fitzgerald made her first tour of Australia in July 1954 for the Australian-based American promoter Lee Gordon.\"\n",
     "posnoised_doc = posnoise_instance.pos_noise(document)\n",
     "\n",
     "print(document)\n",
@@ -73,7 +45,15 @@
   {
    "cell_type": "code",
    "execution_count": null,
-   "id": "6e92a348",
+   "id": "dfa9dad7",
+   "metadata": {},
+   "outputs": [],
+   "source": []
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "b153a880",
    "metadata": {},
    "outputs": [],
    "source": []

README.md

@@ -24,22 +24,39 @@ The latter will pull and install the latest commit from this repository as well
 
 ## Quickstart
 ```python
-import posnoise
 from posnoise import POSNoise
 
-posnoise_instance = POSNoise(spacy_model_size=posnoise.core.SpacyModelSize.Large)
+# By default POSNoise loads the "Large" spaCy model
+posnoise_instance = POSNoise()
+    
+# In case you want specify another model, just set it accordingly e.g. 
+# posnoise_instance = POSNoise(spacy_model_size=posnoise.core.SpacyModelSize.Medium)
 
-document = "I love python !"
+document = "Fitzgerald made her first tour of Australia in July 1954 for the Australian-based American promoter Lee Gordon."
 posnoised_doc = posnoise_instance.pos_noise(document)
 
 print(document)
+# Fitzgerald made her first tour of Australia in July 1954 for the Australian-based American promoter Lee Gordon.
+
 print(posnoised_doc)
- 
-# I love python !
-# I Ø # !
+# § made her first # of § in § µ for the §-Ø @ # § §.
 ```
 
-A detailed explanation of the **POS placeholders** (Ø, #, etc.) used in POSNoise can be found in Table 2 in the paper. Here, the arXiv version for open access: https://arxiv.org/abs/2005.06605
+All part-of-speech (POS) placeholders used in POSNoise to replace topic-related words or tokens are listed below:
+
+| **Category**   | Tag | Examples                                      |
+|----------------|-----|-----------------------------------------------|
+| **Noun**        | #   | { house, music, bird, tree, air, … }         |
+| **Proper noun** | §   | { David, Vivien, London, USA, COVID-19, … }  |
+| **Verb**        | Ø   | { eat, laugh, dance, travel, hiking, … }     |
+| **Adjective**   | @   | { red, shiny, fascinating, phenomenal, … }   |
+| **Adverb**      | ©   | { financially, foolishly, angrily, … }       |
+| **Numeral**     | μ   | { 0, 5, 2013, 3.14159, III, IV, MMXIV, … }   |
+| **Symbol**      | $   | { £, ©, §, %, #, … }                         |
+| **Other**       | ¥   | { xfgh, pdl, jklw, … }                       |
+
+
+_Source_: [POSNoise: An Effective Countermeasure Against Topic Biases in Authorship Analysis](https://arxiv.org/abs/2005.06605) ➜ Table 2. 
 
 ## Features
 - Effective way to mask topic-relatd content with **custom POS placeholders**

posnoise/core.py

@@ -118,7 +118,7 @@ def __init__(
         self,
         nlp_model: Optional["spacy.language.Language"] = None,
         abbrev_pos_tags: Optional[dict] = None,
-        spacy_model_size: Union[SpacyModelSize, str] = SpacyModelSize.Small,
+        spacy_model_size: Union[SpacyModelSize, str] = SpacyModelSize.Large,
         disable: Iterable[str] = ("parser", "ner"),
         verbose: bool = False,
         log_fn: Optional[Callable[[str], None]] = None,