As the experienced senior petroleum engineers retire, new engineers must learn experience from millions of pages of documents accumulated for more than 10 years in a short period. For the oil industry that counts every second, the cost of delayed exploitation is astonishing. It will be a time-consuming process for each new engineer to learn from the huge amount of documents and files, and to make the correct decisions.
IBM Watson imports a total of 38,000 documents recording the practical engineering experience of Woodside over the past 30 years, into Watson Explorer, which also builds the enterprise KM assistant named Willow Virtual Search Assistant. In this way, the engineers could quickly find relevant data history from millions of files, so as to make the correct decisions.
1.Use Watson Explorer to create and train a professional corpus, so as to link up with the Watson API functions
It uses Watson Explorer to build an internal petroleum corpus. The training is conducted by many experienced engineers. With the Watson’s function of context learning, it solves the problem of great variety of reports and different terms used by each engineer. Moreover, it takes advantage of the random text search technology, so the engineers could describe the target problem to be searched in their own way. Through the Watson API, it links up with the Natural Language Classifier, Retrieve and Rank, and Conversation.
2.Link up with API functions, use the natural language to make queries in the corpus, and make response automatically through Conversation
It uses the Natural Language Classifier to analyze the intention of the engineer's dialogue, which could understand different expressions, and achieve conservation between the engineer and Willow. After understanding the engineer's problem, Watson collects all relevant information from the Corpus, and analyzes the correlation by using Retrieve and Rank. After rating each response information, it will choose the optimal ones as the answer. Finally, it combines Conversation with artificial voice, and gives response to the engineer vocally.
The recommendation engine is to help recommend the goods that will be possibly purchased based on the member's preferences. Moreover, it reduces the annoying advertising letters for members, achieving a win-win outcome.