[vc_row row_type=”3″ blox_image=”21394″ align_center=”aligncenter” page_title=”page-title-x” blox_dark=”true” parallax_speed=”8″][vc_column]

---

---

[vc_column_text]

Extracting merchant names from bank transaction records

[/vc_column_text]

---

---

[/vc_column][/vc_row][vc_row][vc_column width=”1/4″]

---

---

[/vc_column][vc_column width=”1/2″]

---

[vc_column_text]

Financial transaction records contain transaction descriptions, which encodes some information in it including merchant location, transaction type, merchant name, etc. Because the encoding format of the string is not standardized, it is nearly impossible to write rules-based code to parse it.

The US-based startup approached us to help them with extracting that information (merchant name in particular) from the transaction record.

[/vc_column_text]

---

[/vc_column][vc_column width=”1/4″][/vc_column][/vc_row][vc_row][vc_column width=”1/3″][vc_single_image image=”21395″ img_size=”550×350″][/vc_column][vc_column width=”1/3″][vc_single_image image=”21396″ img_size=”550×350″][/vc_column][vc_column width=”1/3″][vc_single_image image=”21397″ img_size=”550×350″][/vc_column][/vc_row][vc_row][vc_column]

---

[vc_column_text]

About this project

[/vc_column_text][vc_row_inner][vc_column_inner width=”1/2″][vc_column_text]Project description

We have built a Machine Learning model (Convolutional Neural Networks for text in particular) using one million transaction records and different augmented datasets to extract merchant names from the transaction strings. The model worked well even for the merchants that were not seen before.

[/vc_column_text][/vc_column_inner][vc_column_inner width=”1/2″][vc_column_text]Technologies:

• PyTorch based Convolutional Neural Networks for text.

[/vc_column_text][/vc_column_inner][/vc_row_inner]

---

[/vc_column][/vc_row]