Kent Taylor, Acrolinx
Every once in a while an idea or concept or product comes along that radically changes the way we think about and approach different aspects of our lives. Breakthroughs in medical, energy, computing, communication, transportation, and entertainment technologies have had obvious impacts in our personal lives, but what changes have had an impact on our professional lives?
The introduction of the printing press changed the way our ancestors thought about written communication – moving static information from the hands of the privileged few to the masses. For the next 500 years, the technology improved incrementally, but the basic concept remained the same. Then, in 1890, the linotype machine changed the concept of printed communication from static to dynamic – spawning monthly, weekly, then daily printing of high-value information.
Then came the ‘digital revolution’ starting in the 1980s. Digital technology enabled rapid and profound change in the technology of written communication: WYSIWYG, Computer Graphics, On-line Documents, the World Wide Web, Multimedia, Structured writing, SGML, XML, DITA, Content Management, Single Sourcing, Globalization, Translation and Localization, Controlled Language, Simplified English … . In addition to all the changing technology, we’ve also had to deal with ‘right-sizing,’ outsourcing, off-shoring, and unprecedented merger and acquisition activity over the past quarter-century. The digital revolution changed everything.
Well, almost everything… Creating, editing, and measuring the quality of the content that is ultimately distributed to the information consumer is still done the same way it was done in Gutenberg’s time – by human beings. Relatively simple spelling and grammar checkers have been around for awhile, and have no doubt helped to improve the average person’s writing. Unfortunately, these tools come up short when applied to the industrial-strength technical writing environment. But that’s finally changing, too!
Decades of research in Natural Language Processing (NLP), combined with increasingly powerful computers have enabled the development of a new generation of computerized content analysis and editing programs. These sophisticated programs analyze the quality of written text the same way human copy editors do – checking sentence structure, grammar, and spelling, and putting special emphasis on conformance to corporate or industry-mandated style rules and terminology guidelines.
While quality of content has always been important, it has been time consuming and expensive to ensure, and it is often sacrificed when time or cost pressures arise. (Hence the cliché: ‘Cost, Quality, Timeliness – pick two.’) But times are changing. As companies have branched out to global markets, they have seen translation/localization and customer support/call center costs skyrocket. Better quality of original source language content has been shown to stabilize or reduce these costs of going global, and interest in content quality improvement has grown. The concept is simple: clear, concise, consistent, and correct (grammar and spelling) content is easier to understand and easier, quicker, and less costly to translate. It is also more likely to be read, leading to reductions in calls to customer support specialists.
While the concept is simple, implementation is a bit more difficult. Most large corporations have formal style guides and glossaries that encourage quality and consistency, but ensuring conformance to these guidelines requires initial and ongoing training, independent review and editing, and some form of final Quality Control to be effective. Before the introduction of tools like acrocheck, identifying and fixing these conformance issues tended to be labor-intensive, time consuming, and expensive.
There are several tools and packaged processes available today that address parts of the problem, but most come with preconceived notions of style and terminology and require writing and editing staff to change the way they work. We believe that the most effective approach uses existing tools and processes, and disrupts existing workflows as little as possible. This approach will require minimal training by employing intuitive features that
- embody ‘best practices’ spelling, grammar, and (writing) style rules, quickly customizable to company-specific preferences
- provide guidance for writing content that can be easily and cost-effectively translated (for example, avoid noun strings, long sentences, separated verb parts, split infinitives, and other constructs that are known to cause translation difficulties.)
- enable effective management of industry- and company-specific terminology
- provide real-time support and guidance to authoring and editing staff
- provide largely automated content quality checking/copy editing capability for editors, project managers, and quality managers
The acrocheck™ suite of content quality management tools were designed and developed to meet these needs. And our customers agree; acrocheck is the proven tool of choice for many Global 1000 firms that put a high value on the quality of their content. This attention to quality pays financial dividends in the form of
- improved readability, especially among non-native speakers and poor readers
- improved translatability (10% to 25% through out and cost improvements are typical)
- fewer calls to Customer Support
- reduced litigation risk
- productivity improvements throughout the entire content supply chain
The acrocheck Content Quality Management Tool Suite
The acrocheck product is a sophisticated suite of quality management tools developed by acrolinx GmbH, a spin off of the German Research Centre for Artificial Intelligence (DFKI in Saarbrücken, Germany). With over 200 staff-years of R&D built into the underlying Natural Language Process and statistical analysis engines, the acrocheck tool suite is built on a firm foundation. And extensive usability testing plus five years of real-world deployment experience in a wide variety of industries and companies have resulted in an effective user interface that is well accepted by the professional writing community.
The acrocheck tools can be applied anywhere in a company’s content supply chain, from early Marketing Requirements Documents, through Design Record Documents and Support Documents to Marketing/Sales and Customer Documents and Training material. The tools help ensure that corporate publication standards (style, terminology, grammar, and spelling) are maintained across the entire organization. Whether the final content is ultimately presented on paper, on line, or on optical media, acrocheck tools can help to ensure consistent quality.
Employing a centrally managed server for linguistic analysis and reporting, acrocheck client plug-ins are available for common editing tools including Word, Epic, FrameMaker, XMetaL, and AuthorIT. The acrocheck plug-ins provide support for Writers and Editors directly in the information development process. From a writer/editor perspective, acrocheck looks and acts like “spellcheck” – on steroids! Imagine sophisticated spelling and grammar checking, plus detailed terminology and style guide conformance checking, all without learning any new tools, without extensive training, without external references, and with virtually no effort. The writer/editor interface is shown in the following illustration.
New in Version 3.1 of acrocheck is an ‘Intelligent Reuse’ feature. Our sophisticatedlinguistic matching capability provides the ability to identify ‘standard’ phrases and sentences that are frequently used across documents. We then make these segments available for reuse in an innovative way. Whenever a check is run, the system looks for sentences that contain the same thoughts as any predefined standard sentences. Matches are offered as direct drop-in replacements for what the author had written. The system uses a linguistic matching algorithm that correctly identifies candidate sentences with 90%+ precision. The writer can get his or her thoughts down without interruption, then let acrocheck quickly identify reuse possibilities. This feature is illustrated below, where several sentences (green highlights) are correctly identified as candidates to be replaced by the standard sentence presented in the acrocheck dialog box. (The purple highlight is a terminology flag – improper use of a trademarked term.)
It should be evident from this example that acrocheck is not using a fuzzy matching algorithm to identify standard sentence candidates.
In addition to the writer/editor view, acrocheck’s extensive server-based quality metrics and reporting capabilities provide management with both summary and detailed information needed to monitor service levels and optimize workflows. These quality metrics can also be integrated with database, document management, content management and/or workflow systems to further automate information development processes. The following illustration is an example of the acrocheck management interface, displaying quality metrics for a documentation project. (The documents are color coded to reflect corporate quality thresholds, while specific error types and rates are provided in the table.)
The server also provides central management of corporate resources, including style rules, terminology, and quality thresholds. Separate User-License Management and Terminology Lifecycle Management interfaces are also provided. The Terminology interface is available for importing, exporting, creating, editing, and deleting corporate terminology; Version 3.1.1 of the interface provides for entry of rule-based terminology, and automates many of the tasks associated with the entry of deprecated terminology. One view of this interface is illustrated below.
A typical acrocheck deployment is generally done in two-phases. The first phase deploys a standard acrocheck system (including our Best Practices grammar and style rules, and technical professional spelling dictionary) integrated with customer-specific terminology. This initial production environment enables the writing and editing staff to develop an understanding of, and appreciation for the power of the acrocheck solution, while providing immediate improvements in the quality and translatability of the content being developed.
The second phase, beginning 2 to 6 months after the initial deployment of acrocheck, provides for enhancement of the standard linguistic rules to reflect the precise requirements for the final configuration of each customer deployment. These requirements are derived from corporate style guide(s) and feedback from the initial implementation of acrocheck. The final customized system is generally delivered within 2 to 4 weeks of receipt of requirements.
The Bottom Line
Sophisticated XML applications, Document and Content Management Systems, Automated Workflow and Translation Management Systems all have their place in an effective content supply chain – but when it comes to what your employees and customers actually see, it’s what’s between the tags that counts!