Content consistency is long been the brass ring of quality for technical communication departments, and is starting to be recognized as an important part of an overall content strategy throughout the enterprise. Because so much content is customer facing, or translated and distributed in multiple languages, re-used in ways that could change meaning in critical ways, or affect the organization’s brand, the value of quality and consistency needs to be measured.

Now it can be measured. acrolinx has developed acrocheck, the only product of its kind using meaning-based, natural language processing software to provide quality assurance support by checking terminology, style, and grammar for consistency. By searching large bodies of content, from website content to technical manuals to translation memories to software UI strings, acrocheck generally discovers 15% to 25% redundancy, often times more. For companies translating into multiple languages, this means a large impact on translation dollars.

In late 2007, acrolinx launched the beta of a module that supports intelligent re-use of content through a micro-clustering process developed to wring even more efficiency and consistency out of an organization’s content. The software analyzes content at the phrase and sentence levels to look for similar content, then clusters the content together and generates a report showing the clusters, to allow a human to make the final decision about what content variants should be considered redundant and which need to stay.

During the three months of the beta period, the results have ranged from a mere two ways of expressing a concept to over 200 ways. The software uncovered, in one case, that the simple phrase “Turn the switch to the Run position” had many variations within the same organization, from “Turn switch to Run” to “Switch to the Run position” - and 127 other variations! In another example, analyzing GUI strings from a suite of enterprise software found 30 variations on “see your system administrator” and the error message “enter an end date that is later than the start date.” Given that the text was translated into several languages, the company soon realized how much could be gained by standardizing their vocabulary.

An important aspect of the process is that the software is the support system for the user, who never loses control of the content. Once of the content has been clustered and analyzed for best fit - this is done by comparing the style, terminology, and grammatical rules to other content in the style guide - the user is presented with correction options. Once the initial pass is done, the software can continue supporting users on a regular basis, presenting them with options as they create text. Rather than the writer going to the company style guide to look up the official term, phrase, or product name, the style guide comes to the writer.

Kent Taylor, the General Manager of acrolinx North America, notes with interest that the market for acrocheck started with companies wanting to improve efficiencies around translation costs for technical publications by cleaning up the source content. While that continues to be a strong part of their business, Taylor sees an interesting shift to terminology management and re-use. Companies with large volumes of content are starting to see the value of consistency in corporate image: how does their marketing material describe the company, how to enforce standard presentation of logos, tag lines, brand promotion, and so on. Having style guides and trademark rules are fine, but with a fast-paced, geographically-dispersed team, organizations are looking for any reinforcements they can get. Not only does acrocheck provide assisted authoring, it also produces quality control metrics to show where the problems are and where to concentrate quality improvement efforts, easing the management overhead to discover and calculate this.

Learn more about Acrolinx and its flagship product, acrocheck, in the Exhibition Hall at DocTrain West.

Kent Taylor will also present on the topic of Manage Your Messaging with Machine-Assisted Editing and Large Scale Sentence-level Reuse.