The Engineering Company for the Development of Digital Systems
Committed to providing creative solutions and smarter AI technologies over the years
Digitizing content of the historical manuscripts in the British Library to simplify and streamline search and access.
Problem
Searching for books in the British Library requires a lot of time and effort, especially if one is searching for specific words within the pages of books.
Solution
Using Sotoor, we are able to digitize all pages of books and store them in a suitable format (XML), in a fast and efficient manner. Sotoor’s adaptation feature ensures higher quality.
Outcomes
Digitized the content of ~900k pages in XML format, with text content as well as the metadata for each page.
A comprehensive solution that connects the application directly to the scanner, in order to generate reports from the Arabic hand-filled forms within seconds, using Sotoor’s OCR & OMR technologies.
Problem
Manually typing out the results of hand-filled forms requires a lot of time and effort, slowing up the process of decision making.
Solution
Using Sotoor, a custom engine for processing hand-filled forms is built and provided for end-users through a web application with a user-friendly UI.
Outcomes
On-premise software utilizing auto-feeder scanners to scan applications and automatically convert them into digital forms, as well as extract statistics and provide valuable reports.
Digitizing content of the Egyptian Patent Office documents to simplify and streamline search and access of patents.
Problem
To verify the authenticity of newly submitted patents, employees must manually search past patents, which requires hours, or even days.
Solution
Using Sotoor, we are able to digitize all pages of patents and store them in a suitable format (PDF, DOCX, XML) in an efficient and timely manner.
Outcomes
Digitized content of ~300k pages in different formats (word, searchable pdf, xml). The results were highly accurate, as the average character-based accuracy was 95%+.
Optimize the digital content of Sakhr’s archive of magazines (http://archive.alsharekh.org/) to facilitate search and access.
Problem
A huge amount of magazines are available in alsharekh archive, but there is no automatic search option available, which beats the functionality of this archive.
Solution
Using Sotoor, we can digitize all the magazines pages and store them in a suitable format (XML), in an efficient and timely manner. Sotoor’s adaptation feature ensures higher quality.
Outcomes
Digitized content of ~1.8M pages in XML format, with an average character-based accuracy of 95%+.
Digitizing the decisions and equations documents found in SCU so they can be used for semantic search purposes.
Problem
Searching the huge number of decisions & equations documents found in SCU is not easy. Employees have a hard time locating their search results, especially that some terminologies have multiple synonyms.
Solution
Build a data parser that handles the archive of decisions & equations documents as a special database. Also, using a semantic-search module that can retrieve more optimized results.
Outcomes
An on-premise system that can be used by SCU employees to upload the new documents, and search stored records in multiple ways, exact matching, specific fields, or semantically.
Automatically transcribe the speech content of different Arabic media channels instead of doing that manually through a huge data entry team.
Problem
Transcribing daily Arabic media channels speech content requires a lot of time and costs a lot of money when done manually.
Solution
Using Kateb, the Client will be able to transcribe huge amounts of speech content automatically at a lower cost. Kateb’s processing time is at least 5 times faster than manual transcription.
Outcomes
Kateb On-premise. Able to transcribe thousands of speech hours monthly, and storing the data to be used for further analytics tasks later.
A system to assist decision-makers in Saudi judicial systems. Judges and lawyers can search for past cases similar to their search term. They can view the contents of these cases from allegation, defense, reasons, to final judgment. Users can complete their tasks faster and more accurately, based on previous cases available in the system.
Problem
Judges need a long time to search and learn about similar cases to the case in question. This often leads to a delay in cases, some of which may require an urgent decision.
Solution
RDI’s NLP engines can be adapted to the judicial domain, so that they can be used for semantic search, stemming, NER, keywords extraction, case-type classification, and case ranking based on a relevancy score.
Outcomes
A cloud-hosted web-application that can be used easily by users (judges) for searching the cases database, using RDI’s advanced semantic search engine. Also, the web-app provides a dashboard with statistics about archived cases.
