Spaces:
Sleeping
Sleeping
File size: 12,354 Bytes
8ac0465 71ae214 8ac0465 71ae214 8ac0465 71ae214 34b9f16 71ae214 8ac0465 4e7295f 8ac0465 4e7295f 8ac0465 67034c1 42cf510 f58fdc6 67034c1 42cf510 85879ff c927507 f58fdc6 c927507 25ca557 8856719 c927507 25ca557 8856719 42cf510 e60cb43 8ac0465 4e7295f 71ae214 4e7295f 8ac0465 33b86d9 8ac0465 9120b3c 71ae214 8ac0465 a01f519 8ac0465 a01f519 2c036d1 8ac0465 9be362b 53e4d47 5a23a96 71ae214 9be362b 5a23a96 71ae214 8ac0465 33b86d9 71ae214 5a23a96 33b86d9 71ae214 5a23a96 32a5b43 8ac0465 f1d471f 8ac0465 c3a4cc2 2680ab2 4e7295f 4c335f0 4e7295f 71ae214 8ac0465 a01f519 4e7295f |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 |
# This is our main interface library
# For main things
import types
import bibtexparser
import ethnicolr
import genderComputer
import gender_guesser.detector
import nameparser
import numpy
import pandas
import plotly.express
import streamlit
import st_aggrid
class References(object):
def __init__(self, reference_text):
self.gender_options = ['male', 'mostly_male', 'andy', 'mostly_female', "female", "unknown",
"first_name_initial"]
self.gender_results = {key: 0 for key in self.gender_options}
self.race_options = ['pctwhite', 'pctblack', 'pctapi', 'pctaian', 'pct2prace', 'pcthispanic', 'race_unknown']
self.ethnicity_results = {key: 0 for key in self.race_options}
self.raw_results = pandas.DataFrame(columns=["First Name", "Last Name", "Title"])
# Parse names from input
self.reference_text = reference_text
self.references = bibtexparser.loads(reference_text)
for paper in self.references.entries:
if "author" in paper:
authors = paper["author"].split(' and ')
for person in authors:
name = nameparser.HumanName(person)
self.raw_results.loc[len(self.raw_results.index)] = [name.first, name.last, paper['title']]
def infer_ethnicity(self):
if ethnicity_model == "ethnicolr - census data":
other = ethnicolr.pred_census_ln(self.raw_results, 'Last Name', 2010)
self.raw_results['Most Likely Ethnicity'] = other['race']
elif ethnicity_model == "ethnicolr - wikipedia data":
other = ethnicolr.pred_wiki_name(self.raw_results, 'Last Name', 'First Name')
self.raw_results['Most Likely Ethnicity'] = other['race']
self.raw_results.drop(columns=['__name'])
elif ethnicity_model == "ethnicolr - Florida registration data":
other = ethnicolr.pred_fl_reg_name_five_cat(self.raw_results, 'Last Name', 'First Name')
self.raw_results['Most Likely Ethnicity'] = other['race']
self.raw_results.drop(columns=['__name'])
elif ethnicity_model == "ethnicolr - North Carolina data":
other = ethnicolr.pred_nc_reg_name(self.raw_results, 'Last Name', 'First Name')
self.raw_results['Most Likely Ethnicity'] = other['race']
self.raw_results.drop(columns=['__name'])
for i in self.raw_results['Most Likely Ethnicity']:
self.ethnicity_results[i] = self.ethnicity_results.get(i, 0) + 1
def infer_gender(self):
most_likely_gender = []
if gender_model == "gender_guesser":
d = gender_guesser.detector.Detector()
for name in self.raw_results['First Name']:
if (len(name) == 2 and name[1] == '.') or len(name) == 1:
most_likely_gender.append("first_name_initial")
else:
most_likely_gender.append(d.get_gender(name))
elif gender_model == "genderComputer":
gc = genderComputer.GenderComputer()
for idx in self.raw_results.index:
fn = self.raw_results['First Name'][idx]
if (len(fn) == 2 and fn[1] == '.') or len(fn) == 1:
most_likely_gender.append("first_name_initial")
else:
most_likely_gender.append(
gc.resolveGender(self.raw_results['First Name'][idx] + " " + self.raw_results['Last Name'][idx],
None))
if most_likely_gender[-1] is None:
most_likely_gender[-1] = "unknown"
self.raw_results['Most Likely Gender'] = most_likely_gender
for i in most_likely_gender:
self.gender_results[i] = self.gender_results.get(i, 0) + 1
label_to_gender = {'male': "Very Likely Male",
'mostly_male': "Likely Male",
'andy': "Hard to Tell",
'unisex': "Hard to Tell",
'mostly_female': "Likely Female",
"female": "Very Likely Female",
"unknown": "Unknown (model inconclusive)",
"first_name_initial": "Unknown (first name initial only)"}
label_to_ethnicity = {
'white': 'White',
'black': 'Black',
'api': 'Asian or Pacific Islander',
'hispanic': 'Hispanic',
'pctwhite': 'White',
'pctblack': 'Black',
'pctapi': 'Asian or Pacific Islander',
'pctaian': 'American Indian or Alaskan Native',
'pct2prace': 'Two or more races',
'pcthispanic': 'Hispanic',
'race_unknown': 'Unknown (not found in database)'}
ethnicity_to_label = {v: k for k, v in label_to_ethnicity.items()}
gender_to_label = {v: k for k, v in label_to_gender.items()}
def make_table():
if 'table_data' in streamlit.session_state:
df = streamlit.session_state['table_data']
else:
refs = References(streamlit.session_state.bib)
refs.infer_gender()
refs.infer_ethnicity()
df = refs.raw_results
df = df.replace({"Most Likely Ethnicity": label_to_ethnicity})
df = df.replace({"Most Likely Gender": label_to_gender})
df = df.sort_values(["Last Name", "First Name"])
df = df.reset_index(drop=True)
gb = st_aggrid.GridOptionsBuilder.from_dataframe(df)
gb.configure_default_column(editable=True)
gb.configure_column('Most Likely Ethnicity',
cellEditor='agRichSelectCellEditor',
cellEditorParams={'values': list(label_to_ethnicity.values())},
cellEditorPopup=True
)
gb.configure_column('Most Likely Gender',
cellEditor='agRichSelectCellEditor',
cellEditorParams={'values': list(label_to_gender.values())},
cellEditorPopup=True
)
gb.configure_column('Title',
editable=False
)
response = st_aggrid.AgGrid(
data=df,
gridOptions=gb.build(),
fit_columns_on_grid_load=True,
update_mode=st_aggrid.GridUpdateMode.VALUE_CHANGED,
height=400,
)
streamlit.session_state['table_data'] = response.data
# if response.column_state:
# streamlit.experimental_rerun()
# Define a function for addition
def make_results():
data = streamlit.session_state['table_data']
refs = types.SimpleNamespace(
ethnicity_results=data['Most Likely Ethnicity'].value_counts().to_dict(),
gender_results=data['Most Likely Gender'].value_counts().to_dict(),
)
plt1 = plotly.express.pie(
names=list(refs.ethnicity_results.keys()),
values=refs.ethnicity_results.values(),
hole=0.5,
)
plt2 = plotly.express.pie(
names=list(refs.gender_results.keys()),
values=refs.gender_results.values(),
hole=0.5,
)
plt1.update_layout(legend=dict(orientation="h"))
plt2.update_layout(legend=dict(orientation="h"))
col1, col2, col3, col4 = streamlit.columns(4)
col1.metric("Ethnicity Unknown",
"{pct:.0%}".format(pct=len(data[data['Most Likely Ethnicity'].str.contains('known')]) /
len(data))
)
col2.metric("Gender Unknown",
"{pct:.0%}".format(pct=len(data[data['Most Likely Gender'].str.contains('known|Hard')]) /
len(data))
)
col3.metric("G or E Unknown",
"{pct:.0%}".format(pct=len(data[numpy.logical_or(data['Most Likely Gender'].str.contains('known|Hard'),
data['Most Likely Ethnicity'].str.contains(
'known'))]) /
len(data))
)
col4.metric("G and E Unknown",
"{pct:.0%}".format(pct=len(data[numpy.logical_and(data['Most Likely Gender'].str.contains('known|Hard'),
data['Most Likely Ethnicity'].str.contains(
'known'))]) /
len(data))
)
streamlit.plotly_chart(plt1, use_container_width=True)
streamlit.plotly_chart(plt2, use_container_width=True)
streamlit.title("Welcome and thanks for coming")
streamlit.markdown("""Simply put, many people often cite people that are like them. This is a problem because academia has historically been white male dominated, leading to the suppression of marginalized voices. If your citations are biased towards people who look like you, then you are missing out on high-quality work.
Its important to note that using this site is not a replacement for truly being diligent and engaged in citing diverse voices. Rather, this site is just a place to start, and hopefully the first step in your journey of citing more diversely. To learn more about your duty to dismantle institutional oppression through your citation practices, read up here:
- [Cite Black Women](https://www.citeblackwomencollective.org)
- [The Racial Politics of Citation](https://www.insidehighered.com/advice/2018/04/27/racial-exclusions-scholarly-citations-opinion")
- [Inclusive Citation: How Diverse Are Your References?](https://blog.mahabali.me/writing/inclusive-citation-how-diverse-are-your-references/")
""")
streamlit.markdown("To use our tool, copy and paste your references in the box below and click on the "
"`Analyze` button.")
filler = """@article{Raina2019,
author = {Raina, Ayush and McComb, Christopher and Cagan, Jonathan},
title = {Learning to Design From Humans: Imitating Human Designers Through Deep Learning},
journal = {Journal of Mechanical Design},
volume = {141},
number = {11},
year = {2019},
month = {09},
issn = {1050-0472},
doi = {10.1115/1.4044256}
}
@article{Williams2019,
author = {Williams, Glen and Meisel, Nicholas A. and Simpson, Timothy W. and McComb, Christopher},
title = {Design Repository Effectiveness for 3D Convolutional Neural Networks: Application to Additive Manufacturing},
journal = {Journal of Mechanical Design},
volume = {141},
number = {11},
year = {2019},
month = {09},
issn = {1050-0472},
doi = {10.1115/1.4044199}
}"""
if "bib" in streamlit.session_state:
filler = streamlit.session_state["bib"]
streamlit.text_area(".bibtex only for now, sorry!", filler, key="bib", height=250)
details = streamlit.sidebar
details.markdown("## Settings")
details.markdown("The following dropdowns allow you to select the specific models that are used for analysis.")
details.markdown("When it comes to gender inference, you have two models to choose from: [`gender_guesser`](https://github.com/lead-ratings/gender-guesser) (which is primarily trained on European names), and [`genderComputer`](https://github.com/tue-mdse/genderComputer) (which draws on more global data).")
gender_model = details.selectbox("Gender Inference Model", ("gender_guesser", "genderComputer"))
details.markdown("The ethnicity inference models are all drawn from the [`ethnicolr`](https://github.com/appeler/ethnicolr) library. The models are differentiated by the data they are trained on.")
ethnicity_model = details.selectbox("Ethnicity Inference Model", ("ethnicolr - census data",
"ethnicolr - wikipedia data",
"ethnicolr - North Carolina data",
"ethnicolr - Florida registration data"))
placeholder = streamlit.empty()
time_to_analyze = placeholder.button("Analyze")
if time_to_analyze or 'already_analyzed' in streamlit.session_state:
streamlit.session_state['already_analyzed'] = True
placeholder.empty()
with streamlit.spinner("Analyzing..."):
make_table()
streamlit.markdown("These tabs summarize your results with a variety of visualizations and statistics.")
make_results()
|