adding compare demo, style changes

pages/1_🏷_Label_Clause_Demo.py

@@ -57,17 +57,16 @@ def get_prediction_prob(text):
     return y_pred, y_probs
 
 
-st.sidebar.title('Enter clause text')
-text = st.sidebar.text_area(label='', value=EXAMPLE_TEXT, height=250)
+text = st.sidebar.text_area(label='Enter Clause Text', value=EXAMPLE_TEXT, height=250)
 button = st.sidebar.button('**Label Clause**', type='primary', use_container_width=True)
 
-with st.spinner('Loading model...'):
+with st.spinner('⚙️ Loading model...'):
     model = load_model()
 
 classes = [s.upper() for s in model.classes_]
 
 if button:
-    with st.spinner('Processing Clause...'):
+    with st.spinner('⚙️ Processing Clause...'):
         y_pred, y_probs = get_prediction_prob(text)
         explainer = LimeTextExplainer(class_names=[cls[:9] + '…' for cls in model.classes_])
         exp = explainer.explain_instance(text,

pages/2_🏷_Label_Contract_Demo.py

@@ -81,15 +81,17 @@ def get_prediction_prob(text):
     return y_pred, y_probs
 
 
-model = load_model()
+with st.spinner('⚙️ Loading model...'):
+    model = load_model()
+
 classes = [s.title() for s in model.classes_]
 
-text = st.sidebar.text_area('Enter Contract text:', value=EXAMPLE_TEXT, height=500)
+text = st.sidebar.text_area('Enter Contract Text', value=EXAMPLE_TEXT, height=250)
 button = st.sidebar.button('Label Contract', type='primary', use_container_width=True)
 
 if button:
     text = text[:250]
-    with st.spinner('Processing Contract..'):
+    with st.spinner('⚙️ Processing Contract..'):
         y_pred, y_probs = get_prediction_prob(text)
         explainer = LimeTextExplainer(class_names=[cls[:9] + '…' for cls in classes])
         exp = explainer.explain_instance(text,

pages/3_⛏_Extract_Demo.py

@@ -50,10 +50,10 @@ def load_model():
     return nlp
 
 
-text = st.sidebar.text_area('Enter Clause text:', value=EXAMPLE_TEXT, height=350)
+text = st.sidebar.text_area('Enter Clause Text', value=EXAMPLE_TEXT, height=250)
 button = st.sidebar.button('Extract Data', type='primary', use_container_width=True)
 
-with st.spinner('Loading model...'):
+with st.spinner('⚙️ Loading model...'):
     nlp = load_model()
 
 
@@ -133,7 +133,7 @@ def add_rule_based_entites(doc):
 if button:
     col1, col2 = st.columns(2)
 
-    with st.spinner('Extracting Data...'):
+    with st.spinner('⚙️ Extracting Data...'):
         doc = nlp(text)
         doc = add_rule_based_entites(doc)
         doc = add_detected_persons_as_parties(doc)
@@ -165,7 +165,7 @@ if button:
             'Label': []
         }
 
-        st.header('📊 Extracted Information')
+        st.header('📊 Extracted Data')
         for span in doc.spans['sc']:
             data['Label'].append(span.label_)
             data['Text'].append(span.text)

pages/4_🔗_Compare_Demo.py

@@ -0,0 +1,108 @@
+import streamlit as st
+import difflib
+import spacy
+
+
+@st.cache(allow_output_mutation=True)
+def load_model():
+    return spacy.load('en_core_web_md')
+
+
+## Layout stuff
+st.set_page_config(
+    page_title="Compare Demo",
+    page_icon="🔗",
+    layout="wide",
+    initial_sidebar_state="expanded",
+    menu_items={
+        'Get Help': 'mailto:[email protected]',
+        'Report a bug': None,
+        'About': "## This a demo showcasing different Legal AI Actions"
+    }
+)
+
+st.title('🔗 Compare Demo')
+st.write("""
+This demo shows how AI can be used to compare passages of text.
+""")
+st.write("**👈 Enter two passages of text on the left** and hit the button **Compare** to see the demo in action")
+
+with st.spinner('⚙️ Loading model...'):
+    nlp = load_model()
+
+EXAMPLE_TEXT_1 = """This Agreement shall be governed by and interpreted under the laws of the
+State of Delaware without regard to its conflicts of law provisions."""
+
+EXAMPLE_TEXT_2 = """This agreement will be governed by and must be construed in accordance with the laws of the State of Israel."""
+
+text_1 = st.sidebar.text_area('Enter a passage of text', value=EXAMPLE_TEXT_1, height=150, key='input1')
+text_2 = st.sidebar.text_area('Enter a second passage of text', value=EXAMPLE_TEXT_2, height=150, key='input2')
+
+button = st.sidebar.button('Compare', type='primary', use_container_width=True)
+
+
+def get_tokens(doc):
+    return [token.lower for token in doc]
+
+
+def add_md_color(text, match):
+    color = 'green' if match else 'red'
+    return f":{color}[{text}]"
+
+
+def create_str_output(doc, matching_idxs):
+    out = []
+    for token in doc:
+        if any(token.i in range(start, end) for start, end in matching_idxs):
+            match = True
+        else:
+            match = False
+        out.append(add_md_color(token.text, match))
+    return ' '.join(out)
+
+
+if button:
+
+    with st.spinner('⚙️ Comparing Texts...'):
+        doc_1 = nlp(text_1)
+        doc_2 = nlp(text_2)
+
+    st.header('🧪 Comparison')
+    st.markdown('We can highlight the :green[similarities] and :red[differences] across the two texts')
+    col1, col2 = st.columns(2)
+    sm = difflib.SequenceMatcher(None, get_tokens(doc_1), get_tokens(doc_2))
+    matching_blocks = [match for match in sm.get_matching_blocks()]
+
+    doc_1_matching_idxs = []
+    doc_2_matching_idxs = []
+    for a, b, n in matching_blocks:
+        doc_1_matching_idxs.append((a, a + n))
+        doc_2_matching_idxs.append((b, b + n))
+
+    with col1:
+        st.markdown(create_str_output(doc_1, doc_1_matching_idxs))
+    with col2:
+        st.markdown(create_str_output(doc_2, doc_2_matching_idxs))
+
+    col1, col2, col3 = st.columns(3)
+
+    with col1:
+        # perform simple sequence matching
+        sm = difflib.SequenceMatcher(None, get_tokens(doc_1), get_tokens(doc_2))
+        st.subheader('📑 Textual Similarity')
+        st.markdown('We can measure the similarity based on the *wording* of the two texts.')
+        st.metric(label='Textual Similarity', value=f"{sm.ratio() * 100:.1f}%")
+
+    with col2:
+        st.subheader('📏 Linguistic Similarity')
+        st.markdown(
+            'We can measure the similarity based on the *linguistic features* of the two texts.')
+        postags_1 = [token.pos_ for token in doc_1]
+        postags_2 = [token.pos_ for token in doc_2]
+        sm = difflib.SequenceMatcher(None, postags_1, postags_2)
+        st.metric(label='Linguistic Similarity', value=f"{sm.ratio() * 100:.1f}%")
+
+    with col3:
+        st.subheader('💭 Semantic Similarity')
+        st.markdown('We can measure the similarity based on the *meaning* of the two texts.')
+        st.metric(label='Semantic Similarity', value=f"{doc_1.similarity(doc_2) * 100:.1f}%")

pages/5_🗂_Organise_Demo.py

@@ -0,0 +1,17 @@
+import streamlit as st
+
+st.set_page_config(
+    page_title="Organise Demo",
+    page_icon="🗂",
+    layout="wide",
+    initial_sidebar_state="expanded",
+    menu_items={
+        'Get Help': 'mailto:[email protected]',
+        'Report a bug': None,
+        'About': "## This a demo showcasing different Legal AI Actions"
+    }
+)
+
+st.title('🗂 Organise Demo')
+st.write("""This demo shows how AI can be used to organise a collection of texts.""")
+st.header("🏗 This demo is under construction. Please visit back soon.")

pages/6_🔎_Find_Demo.py

@@ -0,0 +1,17 @@
+import streamlit as st
+
+st.set_page_config(
+    page_title="Find Demo",
+    page_icon="🔎",
+    layout="wide",
+    initial_sidebar_state="expanded",
+    menu_items={
+        'Get Help': 'mailto:[email protected]',
+        'Report a bug': None,
+        'About': "## This a demo showcasing different Legal AI Actions"
+    }
+)
+
+st.title('🔎 Find Demo')
+st.write("""This demo shows how AI can be used to find relevant information from a collection of texts.""")
+st.header("🏗 This demo is under construction. Please visit back soon.")

🏡_Home.py

@@ -25,6 +25,8 @@ st.markdown(
     - 🏷 **Label** - Using AI to **label** text
     - ⛏ **Extract** - Using AI to **extract** information from text
     - 🔗 **Compare** - Using AI to **compare** passages of text
+    - 🗂 **Organise** - Using AI to **organise** a collection of texts
+    - 🔎 **Find** - Using AI to **find** relevant information from a collection of texts
 
     **👈 Select a demo from the sidebar** to see some examples of what Legal AI can do!
     
@@ -44,10 +46,10 @@ st.markdown(
     ## About Us
     simplexico offers white-glove Legal AI education, design and development services. We are on a mission to help
     legal professionals step into a future of collaboration with AI. 
+    
     We have the Legal AI Chefs 👩‍🍳 and recipes 📜. 
     You have the ingredients 🥒 (data).
-    
-    Let's work together to tailor some yummy Legal AI meals just for your taste 🥧!
+    We can tailor some yummy Legal AI meals just for your taste 🥧!
     
     ### Want to learn more?
     - 🌐 Check out our website [simplexico.ai](https://simplexico.ai)