Systems and methods for well document extraction
The present disclosure relates to systems and methods for automatically extracting data from well documents. The systems and methods obtain extracted text from the well documents and create a tailored prompt with instructions for extracting the data from the extracted text. The systems and methods provide the tailored prompt to a large language model. The large language model uses the tailored prompt to automatically extract the data from the extracted text. The systems and methods receive the data from the large language model and present the data in an interactive table.
1 . A method, comprising:
obtaining extracted text from well documents comprising a first document having a first format and a second document having a second format, wherein the first format comprises a first unstructured format including at least one of a PNG format or a TIFF format and the second format comprises a second unstructured format including a PDF format;
receiving a tailored prompt with instructions for extracting data from the extracted text;
providing, to a large language model, the tailored prompt;
receiving, from the large language model, the data in response to the large language model using the tailored prompt to automatically extract the data from the extracted text, the data having a third format different than the first format and the second format, the third format comprising a structured text format;
displaying, on a display, the data in an interactive table;
displaying a well schematic using the data, the well schematic comprising quality control parameters;
displaying, on the display, the well schematic nearby the data in the interactive table;
querying the extracted text to obtain actionable document insights;
updating the well documents to include the actionable document insights; and
storing, on a memory, the well documents including the actionable document insights.
2 . The method of claim 1 , wherein the tailored prompt includes an objective to obtain hole casing and top of cement information from the well documents, settings for the large language model, and further instructions for the large language model to use in extracting at least one of hole casing details, inputs, or examples.
3 . The method of claim 2 , wherein:
the further instructions comprise the extracting the hole casing details including guidelines for casing size, casing start depth, casing end depth, casing type, hole size, hole depth, the top of cement, cross-reference information, and output specifications; and
the large language model uses the instructions to automatically extract the data from the extracted text.
4 . The method of claim 3 , wherein the large language model uses the instructions to automatically infer the data from the extracted text.
5 . The method of claim 1 , wherein the tailored prompt includes an objective to obtain well property data from the well documents, sample inputs, examples for the large language model, settings for the large language model, and further instructions for the large language model to use in extracting the well property data.
6 . The method of claim 5 , wherein:
the instructions for extracting the well property data include well name instructions, well American Petroleum Institute (API) number instructions, a unique well identifier (UWI) instructions, and field extraction instructions; and
the large language model uses the instructions to automatically extract the data from the extracted text.
7 . The method of claim 1 , further comprising:
receiving updated data in the interactive table;
automatically displaying a modified well schematic using the updated data; and
presenting, on the display, the modified well schematic.
8 . The method of claim 7 , further comprising:
receiving a determination that the updated data violates an industry standard or a limit of a value by performing quality control on the updated data; and
presenting, on the display, an error message in response to the determination that the updated data violates the industry standard or the limit of the value.
9 . The method of claim 1 , further comprising:
receiving a query for the well documents;
providing, by the large language model, a response to the query using the extracted text; and
presenting, on the display, the response to the query and a citation to a source for the response.
10 . The method of claim 9 , further comprising:
receiving a selection of the citation; and
displaying a well document corresponding to the source nearby the response.
11 . A system, comprising:
a memory; and
a processor operable to communicate with the memory, wherein the processor is operable to:
obtain extracted text from well documents comprising a first document having a first format and a second document having a second format, wherein the first format comprises a first unstructured format including at least one of a PNG format or a TIFF format and the second format comprises a second unstructured format including a PDF format;
receive a tailored prompt with instructions for extracting data from the extracted text;
provide, to a large language model, the tailored prompt;
receive, from the large language model, the data in response to the large language model using the tailored prompt to automatically extract the data from the extracted text, the data having a third format different than the first format and the second format, the third format comprising a structured text format;
display, on a display, the data in an interactive table;
display a well schematic using the data, the well schematic comprising quality control parameters;
display, on the display, the well schematic nearby the data in the interactive table;
query the extracted text to obtain actionable document insights;
update the well documents to include the actionable document insights; and
store, on the memory, the well documents including the actionable document insights.
12 . The system of claim 11 , wherein the tailored prompt includes an objective to obtain hole casing and top of cement information from the well documents, settings for the large language model, and further instructions for the large language model to use in extracting at least one of hole casing details, inputs, or examples.
13 . The system of claim 12 , wherein:
the further instructions comprise the extracting the hole casing details including guidelines for casing size, casing start depth, casing end depth, casing type, hole size, hole depth, the top of cement, cross-reference information, and output specifications; and
the large language model uses the instructions to automatically extract the data from the extracted text.
14 . The system of claim 13 , wherein the large language model uses the instructions to automatically infer the data from the extracted text.
15 . The system of claim 11 , wherein the tailored prompt includes an objective to obtain well property data from the well documents, sample inputs, examples for the large language model, settings for the large language model, and further instructions for the large language model to use in extracting the well property data.
16 . The system of claim 15 , wherein;
the instructions for extracting the well property data include well name instructions, well American Petroleum Institute (API) number instructions, a unique well identifier (UWI) instructions, and field extraction instructions; and
the large language model uses the instructions to automatically extract the data from the extracted text.
17 . The system of claim 11 , wherein the processor is further operable to:
receive updated data in the interactive table;
automatically display a modified well schematic using the updated data; and
present, on the display, the modified well schematic.
18 . The system of claim 17 , wherein the processor is further operable to:
receive a determination that the updated data violates an industry standard or a limit of a value by performing quality control on the updated data; and
present, on the display, an error message in response to the determination that the updated data violates the industry standard or the limit of the value.
19 . The system of claim 11 , wherein the processor is further operable to:
receive a query for the well documents;
provide, by the large language model, a response to the query using the extracted text; and
present, on the display, the response to the query and a citation to a source for the response.
20 . The system of claim 19 , wherein the processor is further operable to:
receive a selection of the citation; and
display a well document corresponding to the source nearby the response.
21 . The method of claim 1 , further comprising displaying a graphical user interface including:
a first tab including information regarding automatic wellbore extraction; and
a second tab next the first tab, the second tab showing the actionable document insights in response to the querying the extracted text and a query used in the querying the extracted text.