Click for next page ( 26


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 25
26 Why Create a Knowledge Base? This knowledge base staff can perform other important duties in conjunction with the release or analysis of data. The best argument in favor of a select group of professionals These duties include calculating quality control measures, to support linkages and analyses is that it will provide consis- processing data dictionaries and coding manuals, creating tent and trained service for decision makers by delivering data basic standardized reports, updating contact information, and and analytic support. This type of argument would require documenting the availability of the data. All of these prod- that multiple agencies compare the cost of their analytic sup- ucts support the authorized users, including the custodial port and data sharing efforts at this time and estimate what agencies. they could relinquish if a single knowledge-based staff took on selected tasks. By virtue of their experience working with each of the data sources, the staff acquires knowledge of the contents A likely impetus for creation of such a knowledge base and limitations of each database. They become ideally suited comes from the inability of agencies to meet the rising to serve as an analytic resource to assist other users, and they demand for access to data and requests for assistance in using can communicate the caveats about each database and help the data. In an era of budget constraints and contraction of users formulate questions in a manner that can be reliably agency services to only the "core" business missions, a cen- answered using the data. When users need help conducting tralized knowledge base is one way to gain efficiency and pro- an analysis, the staff can explain the coding and structure of vide a necessary service that normally would be outside of each file or they can conduct analyses for the users. any single agency's purview. It is also a way to eliminate the nonconstructive arguments between agencies about the "offi- Ancillary Benefits of a Knowledge Base cial" numbers on traffic-related injuries, deaths, and costs because everyone would be working from the same set of A knowledge base, whether through a data clearinghouse numbers. Communication about the data and how to use the method or through a specialized staff method, brings the pos- data correctly is also simplified when there is a single source. sibility of meeting users' needs in a highly visible manner. It is assumed that better customer service will support expanded use of the data by a larger number of people drawn from a How a Knowledge Base Works broader spectrum of users. To the extent that these new users see themselves as stakeholders in the traffic records system, On a periodic basis agreed to by each agency that acts as cus- they can in turn support improvements to all the components todian of a component of the traffic records system, a copy of the traffic records system. One important possible benefit of their data is forwarded to the data clearinghouse or the of a traffic records knowledge base is the building of a coali- staff resources assigned to be the knowledge base for safety tion that will help to support expansion and improvement linkage and analysis. The agency will also provide current of the system that supports them. With this in mind, it is sug- documentation, a data dictionary, and contact information. gested that the knowledge base include customer service Ideally, the data submitted are as complete as possible in that measurements and that the staff maintain a customer contact personal identifiers are left in the file at this point to assist in database. linkages. However, the staff would be responsible for remov- ing any identifiers based on the custodial agency's require- ments before releasing any information. Because files may SUMMARY contain information that cannot be released to the public, archival copies are generally not made available to anyone Data collection, data management, and data usage are the except those authorized by the custodial agency and only for three areas that define the success of a crash records system. limited purposes. There is currently no single system in the United States that would meet any reasonable definition of a best practice The staff will generate an extract for general release that approach to all three areas simultaneously. There are, how- has all personal identifying information redacted in accor- ever, examples of systems that are successful in one area or dance with applicable privacy laws. At that point, the data another. Based on these examples, and using the literature will be available for general release to authorized users. The and consultant experience with traffic records systems, some staff can then use the original data again to create one or more overall descriptions of systems that are possible with today's linked databases as needed. Because linkages among traffic technology and could serve the needs of stakeholders at all records components often are most reliable when using per- levels of the traffic safety community were developed. These descriptions are summarized here. sonal identifiers as linking variables, the personal informa- tion is left in the files until the linkage is performed. When this is not possible, or where that linkage fails, a probabilis- Crash Data Collection tic matching process may be needed. The resulting linked datasets can then be purged of personal identifiers and made The most promising approach to crash data collection is an available to authorized users. automated field data collection tool that is used to capture

OCR for page 25
27 information as close to the event as possible. Field data col- data collection and electronic data transfer. The savings in lection hardware can include a portable or in-vehicle com- reduced data entry, along with improvements in data quality puter, GPS unit, magnetic strip and/or bar code reader, and and timeliness, will benefit all stakeholders. other technology as desired. The officer using this tool would be able to link with the state driver and vehicle data to com- plete sections of the crash report without having to reenter Data Linkages for Reporting and Analysis information that already exists electronically. Officers may Crash data alone do not serve as the sole basis upon which to also scan information directly from vehicle identification make highway and traffic safety decisions. A comprehensive numbers and/or registration documents, license plates, or driv- traffic records system is required with linkable components er's licenses to obtain information for their reports. to support analyses of all types of data. In most states, a full traffic records system could not exist in a single agency and The field software tool can include edit checks that match have it fit well with the core business of that agency. For those in the statewide crash report system and prompt offi- example, an agency that is responsible for issuing licenses to cers to complete all required fields, including supplemental drivers and titles to vehicles may not have the resources to reports. A supervisor could then automatically review the support other components of a traffic records system that do resulting crash report. Once accepted, the report can be sent not assist them in completing their agency's primary mis- to the agency's local records management system, if desired, sion. A traffic records knowledge base, either through a data as well as to the statewide crash records system. This paper- clearinghouse or through resources dedicated specifically for less process would also support generation of a graphic ad hoc data linkages and analyses, is a method for a state to image of the form suitable for printing and archival storage. achieve the goal of serving the needs of all highway and traf- The primary advantage of automated field data collection fic safety stakeholders. software is a reduction in the time spent in records manage- ment and supervisory review. The improvement in quality A knowledge base supports all or most components of the and timeliness of the crash data benefits all stakeholders in traffic records system readily available to the users for analy- the traffic safety community. sis and reporting. Data sources are linked directly with the crash data or linked indirectly through probabilistic matching. Crash Data Management This type of knowledge base is one way to increase the util- ity of crash data for less experienced users and to help build Crash records systems must have the capability to accept data strong advocates for traffic records improvement throughout electronically. Adding this capability often results in major the state. The Missouri DOT is an example of a directly updates to the structure and processing of a statewide crash linked data system that primarily supports only that agency's database. However, the document management and archival users. The Massachusetts data warehouse is an example of a storage of crash reports must accommodate both electronic university-based system with Internet access for analysis and and paper forms if the system does not create an electronic reporting given to all approved users. Although the number image of the crash report form. Some manual post-processing of traffic records data clearinghouses is increasing, most of crash information, especially for quality control of loca- states conduct data linkages on an ad hoc basis, often using tion coding, is advisable even with automation of the field university-based staff.