H-Hotels Gruppe Platform search and automatic content linking

Making quick and easy room reservations through the H-Hotels website is one of the key digital assets for the company’s business model. All bookings are made using an infrastructure developed by the H-Hotels Group itself. As H-Hotels are also keen to use innovative technology, they see themselves as technology leaders in the hospitality industry. Together with inovex, H-Hotels have now set themselves the goal of automatically making the existing website content searchable using the latest search technologies in order to further enhance the platform’s usability for customers.

There is a great deal of content available on the H-Hotels website that is not yet used ideally for the user experience and cross-selling. In addition to the instantly visible hotels and offers, these may be ready-made travel packages or hotels that are sold in packages, for example fan packages for football events. This existing content should be searchable, so that hotels matching the recommendations can automatically be displayed to the user. This means that if the website shows content about a short break, suitable hotels should also be displayed instantly. This idea gave rise to the challenge of linking the website content to relevant hotels using natural language processing. The first step was to link hotels to the cities mentioned in the text using text pattern recognition.

The H-Hotels.com website is produced on the basis of the “Neos” open-source content management system and stored as a tree structure in a relational database. Configurable import services periodically collect new content from the site, process it, place it in Elasticsearch, and enrich it with booking data from its own booking engine. The import services can be configured via an HTTP interface. Relevant hotels are recognised in the texts through semi-automatic regular expressions and stored in addition to the imported content. Full search functions could be developed as a result. These include features such as a faceted search (“All hotels from € x to € y”), spelling correction (“Did you mean …”) and phonetic search (“Hamburch” finds content with “Hamburg”). These search functions are also available via an HTTP interface. The import and search functions have been developed in separate applications for better maintainability.

inovex has implemented both the applications described above and modelled the data management in Elasticsearch. Joint workshops on search technologies were conducted to familiarise the H-Hotels developers with the solutions developed by inovex.

One resulting side effect of the project found that the entire H-Hotels.com content is available through the web interface, which allows this content to be integrated into other systems without any manual intervention. For example, the self-developed native H-Hotels app uses this interface to present the current offers, descriptions and much more. The editors at H-Hotels can therefore maintain and update all the e-commerce content for the web and app channels using their existing content management system, which, in addition to avoiding a duplication of work and optimising internal processes, leads to significant cost savings in app development as a separate management interface does not have to be created for the app. inovex delivered the search backend as part of the project, which included the first case study: searching content on the website and automatically linking it to content. The implementation can also be used for a second case study: H-Hotels are now also using the search feature to display content in the mobile app.