Patent Application
20090265364
1. Field of the Invention
This invention relates generally to a method for assigning symptom codes to textual descriptions in reports generated from vehicle service and, more particularly, to a method for assigning symptom codes to textual descriptions in reports generated from service of a vehicle that are analyzed for patterns and relationships.
2. Discussion of the Related Art
When a vehicle has a mechanical, electrical or other problem, the vehicle owner will typically take the vehicle to a service provider, such as a dealership. At the service location, the vehicle owner will describe to a service advisor the problem that the vehicle is experiencing that needs to be fixed. That description is typically typed into a computer, and may include certain diagnostic or trouble codes associated with vehicle parts or procedures. The description is then printed onto a work order, and the work order is given to a technician who will attempt to fix the problem that has been described. As part of the technician's procedure to fix the vehicle, he will also provide a textual description on a service report as to the problems that he sees with the vehicle, the cause of the problems and the operations that were taken to correct the problem. This information is then typed into a computer by an administrative person.
The work order and the service report may then be transmitted to a central database for a particular vehicle manufacturer who is interested in the types of service that have been performed on their vehicles, especially warranty service. This information is important to the vehicle manufacturer because it can be used to increase vehicle quality and consumer satisfaction. This information is available to engineers and other employees of the vehicle manufacturer that may want to determine why a particular problem is occurring, where the engineer can look at the trouble codes and read the text provided by both the service advisor and the technician.
Currently, about 80% of the information on these types of service reports is textual. The most vital diagnostic information is often captured in the text, but the text is difficult to use for decision support activities for various reasons, such as the text being ungrammatical. Available methods for capturing knowledge from text do not work quickly or accurately enough for an early warning or problem solving environment. However, the large amount of diagnostic relevant information, such as repair logs, manufacturing process control documents, launch diaries, etc., can greatly assist in identifying problems before they become overly serious.
In accordance with the teachings of the present invention, a system and method is disclosed for converting text related to vehicle service to symptom codes. The method includes typing into work orders and service reports statements that describe the various symptoms and problems of a vehicle that is being serviced. The work orders and service reports are then transmitted to a database facility where they are analyzed. Prior to the reports being analyzed, the text in the work order and service reports is read by a machine reader that converts the text to symptom codes that describe particular vehicle conditions and symptoms. A processor analyzes the codes for patterns and other relationships, and can provide a display of such patterns. Further, the codes and reports are stored in a memory.
Additional features of the present invention will become apparent from the following description and appended claims, taken in conjunction with the accompanying drawings.
The following discussion of the embodiments of the invention directed to a system and method for converting the text in vehicle service reports to symptom codes is merely exemplary in nature, and is in no way intended to limit the invention or its applications or uses.
As discussed above, the text in the various service reports that are provided by a vehicle service center to a database facility operated by a vehicle manufacturer can only be accessed by persons that actually read the text. The present invention proposes a system and method for converting that text to symptom codes that can then be analyzed by a processor in combination with other diagnostic trouble codes to identify problem areas as quickly as possible. Focusing on symptom descriptions, such as customer complaints, allows a meaningful template for information extraction, overcoming difficulties of ungrammatical text and the ambiguity of general text.
To be useful for decision support, the text data should be mapped to a set of features that ideally have only a finite number of known values. This mapping process can be referred to as information extraction. One characteristic of this type of mapping is that it disambiguates the text by focusing its information into certain restricted channels of maximum interest. There are many possible elements of the disambiguation process. For example, the use of domain-specific knowledge is a key feature of the process. Prognostics/diagnostics text in the automotive industry has a typical structure, and in particular the symptoms given in customer complaints are described according to certain typical formats. For example, many symptoms have the format “part inoperative.” It turns out that identifying part name categories referenced in the text is a core challenge. Part names can be recognized and classified using unique technologies that exploit existing protocols to mimic the leverage of ontologies and use the power of entropy estimates to pick up the most likely interpretations. The part name classification problem is the most difficult part of the process. The other symptoms can be classified using similar ideas. The classification process identifies nodes in the ontologies that most disambiguates the symptom references. These nodes can be used to provide smart indexes for the text, and enable the applications noted.
The information that is put into the computers 12 and 14 is transmitted or sent by some suitable process to a database facility 16 that receives such reports from many service facilities. The particular reports are sent to a machine reading processor 18 that has been programmed to convert the text in the documents provided in by the service personal and technicians to a particular symptom code depending on the text. For example, a particular problem being serviced might be steering wheel pulls to the left. Machine readers that read text are well known to those skilled in the art. The machine reading processor 18 would be programmed with a code for that symptom so that the text is converted to a decipherable context. Thus, the various diagnostic trouble codes and symptom codes in the reports are now provided as computer readable codes, and can be analyzed by a processor 20. The processor 20 can display the codes on a display 22 and can analyze the codes to identify various patterns and relationships between the codes that might be of interest. Further, the codes are stored in a memory 24.
The foregoing discussion discloses and describes merely exemplary embodiments of the present invention. One skilled in the art will readily recognize from such discussion and from the accompanying drawings and claims that various changes, modifications and variations can be made therein without departing from the spirit and scope of the invention as defined in the following claims.
