Doxie.AI

Unlock History!

Historical records, whether in special collections of libraries or basements of companies, have a wealth of information waiting to be extracted. However, mere digitization is not enough, data has to be extracted in a structured manner. Doxie is unlocking this rich data source by building custom pipelines to extract data from the most challenging data sources.

♦

Our Services

✻

Custom Pipelines for Document Transcription through state of the art Machine Learning AI.
Extract Information, Form Structure, and Enable Data Analysis.

Domain Knowledge

We work closely with your domain experts to extract out key values.

Doxie Document Framework

We apply our document framework, pipeline process and build customize machine learning models for your specific use cases.

Desire Output

We delivery custom output format ready to be used in your research.

Our company elevates digitization to the next level and expedites your research!

Get results in weeks Not Years!

✻

Our Projects

♦

Essig Entomology Use Case

Named Entities from Mixed Medium Data

We used machine learning to detect the irregular shape and sizes for insect labels, transcribe valuable data such as collector, location, id, species etc

Field	Value
ID	UC Berkeley EMEC 454458
Collector	R.L Langston
Location	Ore: Harney Co. Frenchglen, 9mi.E.
Date	VII-21-65
Species	Hesperia Comman Harpalus

Index Card - Natural Language Processing

Extract Key Value Pairs [OCR & NLP - entity]

We utilize machine learning to perform image preprocessing, image-to-text prediction, text processing, and incorporate Natural Language Processing Model extract out data to correct fields to put context to words.

Field	Value
Subject	ANNEXATION - British Columbia - Victoria
Publication	Washington Standard
Title	Large meeting held at Victoria at which resolutions were adopted asking the home government to cede the colonies of Vancouver Island and British Columbia to the United States.
Date	October 6, 1866
Pages	p.2, col.2

Menu Extraction

Field	Value
Company	Fredick & Nelson
Item_1	THE CLOWN SUGGESTS
Item_1_Description	Hot Dog in a Bun Milk or Chocolate Milk
Item_1_cost	45
.....	.......

Field	Value
File No.	65930
LastName	GUNTHER
Army Serial No.	1779820
.....	.......

Different Handwritten documents

Transcribe handwritten format into modern fonts text for easier search and processing.

Mix Media Forms

Complex Mix Media Forms

Extract valuable data and map into searchable fields

[image credit: Essig Museum, University Washington Library, National Archives]

[Note: Doxie.AI has been partnering with UC Berkeley's Bancroft Library to incorporate machine learning technology into their digitization project. This work was recently shared at a Bancroft Roundtable series.]

Take the next step

♦

We are keen on unlocking values and discover key data from your collections! We customize our machine learning model to suit our customers' use cases. We work side-by-side with our customer.

Talk to one of our AI experts what Doxie products can do.