The Business Need
A major global shipbroking services provider wanted their communication with on-port agencies processed over 5 different cargo categories. The shipping brokerage company had to rapidly process emails sent by on-field agents at the docks and deliver them as business insights to vessel and cargo owners and operators for smooth vessel and cargo allocation.
Since the vessel-cargo allocation happened at lightning speeds and the one who received the vessel/cargo information first had a competitive edge that helped in cargo allocation negotiations, manual ways of mail interpretation were found to be tedious and time-consuming. Hence Mobius was sought by the company to give an automated way of extracting the mandatory data points from the unstructured data input.
Challenges we faced
The shipbrokers had about 20K emails per day across categories that had to be deciphered and sent to the shipping brokerage company in an easily consumable format within 10 minutes of mail arrival.
Information to be extracted from the emails included critical cargo and vessel attributes. Data incompleteness was also a challenge as not all the emails had all the details in the expected attribute naming structure and mismatched fields had to be interpreted properly and mapped to their corresponding template within the stipulated time.
How we solved the problem
A custom tool was built by Mobius to automate and expedite the entire workflow of email extraction.
Once the agent emails were delivered to our well-secured FTP server, the tool initially prioritizes the huge inflow of emails as per the time of arrival to ensure that mail processing happens without any time lag.
The prioritized emails were then sorted and categorized by the tool as vessel position emails and order emails based on the email subject.
After email classification, both critical and non-critical data points were extracted, with precedence given to the critical information as they drive the end user’s decision to complete a vessel-cargo or vice versa match.
A quality benchmarking process is carried out by experts fortnightly to determine the accuracy of the extracted data. Improvements were made to the extraction rules based on the findings uncovered in the benchmarking process.
As challenging as it was to make sense out of the mounds of the emails in near real-time, Mobius achieved 100% automation in processing the shipping brokerage emails. With periodical benchmarking of the extracted output, an accuracy of 85% was achieved surpassing other automated solutions offered in the market.