Automatic text generation – a fast and efficient way to generate customized texts
The new retail frontier is all about smart voice assistants such as Amazon Echo, Google Home and Co. and self-learning and intelligent chatbots. We’re surrounded by them. They are transforming the customer experience, significantly raising customers’ expectations and increasing demands on Product Information Management.
Already, a range of efficient products and solutions is available to help retailers and manufacturers automate workflows with the help of artificial intelligence, thus accelerating digital transformation. Today much of the content published online is already generated at lightning speed by computers from data and figures. The increasing amount of product information requires a high degree of automation. Providing the necessary up-to-date information at any time and in multiple languages is one of the biggest challenges of marketing automation.
With NLG Cloud, AX Semantics offers an innovative natural language generation (NLG) software solution to this issue. It automatically generates high-quality customized texts. The successful provider aexea assures that the artificial intelligence-based text engine will catapult marketing automation to the next level – internationally, cost-efficiently and tailored to the needs of different audiences. Our consultants have delved into this new, exciting topic and got to the bottom of this technology.
Why automate text generation?
- Content can be generated within seconds. You don’t have to wait for editors and translators. Websites and sales channels are updated in no time and without great effort.
- Good product data sells better. Texts have to be informative, optimized in terms of SEO and quickly available in order to increase conversion rate of your web shop. You won’t be able to achieve this manually in a reasonable time.
- Your content will be available in multiple languages. Since every language is generated separately, the translation process can be streamlined and the corresponding costs can be significantly reduced. This way, you can easily publish content in different languages and open up new opportunities on an international level.
- Automatically generating content through NLG software technology is very cost-effective. E-commerce companies can quickly and efficiently create large quantities of text and thus greatly enhance their offer.
What happens in a text engine?
In order to generate content, the text engine retrieves data from your database. To interpret this data, customized rules that reflect the specific wording and style used in your company, as well as the intended structure of the text are integrated into the system.
The software then follows a four-step approach:
Data analysis
Information is extracted from certain data records and assigned to specific values: Which field indicates whether a product is a TV or a car? Which field contains the value of the screen diagonal? An abstract image of the data record is created.
Data interpretation
The abstract image can be compared to other, similar data records and other facts may be deduced: Is this particular TV large or small compared to other TVs in stock? What can I infer from size, refresh rate and resolution?
Aggregation
The text engine then browses through the vocabulary, the predefined sentence components and the predicates in your set of rules and chooses the ones that match the given information. While all the previous steps were language independent, at this stage a specific language is selected.
Text generation
In the last step the actual sentences are generated based on the translations of single words, the predefined sentence structures and the grammar of the language. Adjectives, verbs and articles are inflected to agree with the head noun, taking into account all exceptions in the target language.
Thus, the engine creates a text that describes and classifies a data record. The final text is then provided via API as feed or as xls-export and can either be automatically published online or be checked manually by an editor before publishing. Each produced content item is unique, even if several texts are rendered using the same object.
PDF Example text: AX Semantics LED TV
Successfully automating content
Well-structured and high-quality data is an essential prerequisite to automate content generation. Thus, a clear MDM and PIM strategy is crucial to success.
parsionate can support your company with the integration of AX semantics. We are there to advise you – from project management of your marketing automation strategy process to the configuration of the appropriate software solution and up to the optimization of the text engine. Our experts have in-depth knowledge of the key factors that affect the successful integration of NLG software technology. We would gladly advise you which modules are relevant for your company and which are critical for the success of your project.
AX Semantics NLG Cloud is a software product provided by aexea GmbH, one of the world’s leading companies in the field of natural language generation. The Stuttgart-based company combines years of experience in journalism and editing with state-of-the-art technology. With its markup language ATML3 (Automated Text Markup Language), AX Semantics offers its users the opportunity to train artificial intelligence to create their own story.
Do you have any questions conserning automatic text generation? Let's discuss!