Fuzzy string matching of companies' names

About

This application provides help in finding legally associated companies.

Input:

• company name, whose related companies we would like to find (query)
• list of companies' names, among which the search is conducted (database)

Output:

• list of potentially associated companies

Prerequisites:

• Python, version >=3.6
• Unix shell (for dependency management)
• API key

API key is required to get access to Google Knowledge Graph Search API.
To get your personal API key, visit: https://developers.google.com/places/web-service/get-api-key. Then save your API key in data/.api_key file.

Setup:

To create virtual environment .env_fuzzy (with all necessary packages) and activate it, run in Unix shell:

source setup.sh

Once .env_fuzzy is created, the same command can be used to activate this virtual environment.

Making queries:

To make queries, specify 2 parameters:

• path to a .csv file with companies' name, using argument -d <database>.csv
• list of queries, using -q <query> or -f <queryfile> arguments (or even both)

Queries in <queryfile> should be separated by newlines.

Query examples:

python launcher.py -d <database>.csv -q <some_company_1> -q <some_company_2>

python launcher.py -d <database>.csv -f <queryfile>

Header

By default, <database>.csv is considered to have no header. Otherwise, use -e, --header argument:

python launcher.py -d <database>.csv -q <company_name> -e

Column choice

By default, first column of <database>.csv is considered to be a list of companies' names. To pick another column, specify one of the following:

• column index (zero-based numbering), using -i, --iloc argument:

python launcher.py -d <database>.csv -q <company_name> -i 2

• or column name, using -l, --loc argument:

python launcher.py -d <database>.csv -q <company_name> -l "company names"

If --loc is specified, --header can be omitted.

Output to file

By default, query results are written to stdout. To write them to a .csv file, use -o, --output argument:

python launcher.py -d <database>.csv -q <company_name> -o <results>

One neat way to load results from <results>.csv is by using pandas:

import pandas as pd

df = pd.read_csv('<results>.csv', index_col=[0, 1])

This will create a pandas.DataFrame with MultiIndex, whose first index is a query name. Then to pick a part of the results, related to one specific query <query>, use:

query_df = df.loc[<query>]

Help

Use -h argument to access help menu:

>python launcher.py -h

usage: launcher.py [-h] -d DATABASE [-q QUERY] [-f QUERYFILE]
                   [-i ILOC | -l LOC] [-e] [-o OUTPUT] [-t THRESHOLD]
                   [-s SHOWEXTRA]

optional arguments:
  -h, --help            show this help message and exit
  -d DATABASE, --database DATABASE
                        provide path to file with database
  -q QUERY, --query QUERY
                        add search query
  -f QUERYFILE, --file QUERYFILE
                        provide path to file with queries
  -i ILOC, --iloc ILOC  provide number of column with company names
  -l LOC, --loc LOC     provide name of column with company names
  -e, --header          if specified: first line in datafile is treated as
                        header
  -o OUTPUT, --output OUTPUT
                        provide path to output file (stdout by default)
  -t THRESHOLD, --threshold THRESHOLD
                        assign threshold value [0, 100] (80 by default)
  -s SHOWEXTRA, --showextra SHOWEXTRA
                        assign how many extra companies whose score is less
                        than threshold are shown (5 by default)