Skip to main content

Searching for better UX: Bringing web style search to the TV

Quick, accurate and easily executable search has become ubiquitous. People are used to reaching for their phone or laptop to look up all sorts of different things, from products to advice, so when searches provide inaccurate results, or when they require further clarification, it becomes a frustrating and clunky experience.

In the context of proliferating channels and shows, the need for search technology is one that content providers are becoming increasingly aware of and must act upon to avoid being left behind by other industries.

Furthermore, operators that can put in place a highly accurate search function can expect to boost customer happiness and spend – a crucial factor in today’s world of cut-throat competition.

Follow the data trail

The techniques to analyse data and create compelling search platforms that resonate with what we really mean when we search for things have been around for some time. In 2012, Google announced its “Knowledge Graph”, which was designed to understand that keywords weren’t just strings of characters but that they referred to real things in the world that are related to each other in meaningful ways.

In 2013, Facebook revealed “Graph Search”, which trawls for results based on the searcher’s friends, content and relationships, as well as wider trends on the site.

Unlike a traditional database, a graph is much more scalable and flexible because it allows the connection of all sorts of (possibly) unexpected kinds of information or records, without the reliance on "tables". This is perfect for content, which often features links that are based around very human traits, for example nicknames for sports teams which basic search functions can easily miss or not understand. The Knowledge Graph method of search helps in discovering content by using search terms that are popularly used to refer to named entities in the content even when those terms are absent in the content item. For example, “Gunners vs. Black Cats” will help bring up the game “Arsenal vs. Sunderland”.

Crucially, search needs to be semantic, not lexical, as this mirrors the way that people process and search for things. For example searching for “Arnie movies” rather than “Arnold Schwarzenegger movies” is a much easier input for the user, and if the same results are provided for both queries, then that’s a better experience.

By collating information about content – such as names of cast, soundtrack and pictures – and feeding this information into a graph search platform to link this together, a complex picture of connections can be created that mirrors the way that humans think and search for content. Think over a hundred million keyword connections to create a truly seamless experience.

Of course, this vast amount of data needs human oversight to ensure that nuanced links are accurate and haven’t been misinterpreted. By linking to existing editorially curated metadata which provides detailed synopses of content, the machine learning capabilities offered by the knowledge graph can bring searching for content fully into the modern era.

Personalising the search experience

Using a knowledge graph can certainly provide a huge boost to the customer experience of search, but what if searches can be pre-empted by personalising TV experiences to such an extent that individuals are presented with content that they might be interested in?

Personalised services are a common feature for brands across many sectors, ranging from retail through to banking. In the TV industry, personalisation is very much a hot topic, with operators across the globe competing to offer customers a hyper personalised viewing experience which curates and recommends shows. There is no doubt that this is important as the vast increase in channels over recent years has the potential to leave customers confused and may even lead to bad experiences.

What, then, makes a good personalisation experience? Brands have experimented with means of self-selection, with customers providing input on particular genres of content which they enjoy, or simple “tick or cross” feedback which allows them to respond to particular suggestions. Graph search is again the key to unlocking a meaningful personalisation experience for viewers.   Working with the knowledge graph and in a similar fashion, the personal graph can tailor results to the individual to strikingly improve performance. Based on statistical machine learning, the personal graph can immediately begin to understand a user’s preferences. Searched for “City game” and chose Manchester City rather than Norwich City? Next time, Manchester City games can be actively recommended. Is it early in the morning? Then it’s time to recommend a news channel rather than films, based on past user activity. Similarly, trending content or topics can also be prioritised to ensure relevance in results, such as finals of sports competitions or major breaking news.

Graph based search can also identify micro-genres from the connections present in the Knowledge Graph that the viewer may not even be aware of themselves and build on these to provide more detailed recommendations. For example, a viewer may show preferences for items related to a micro-genre such as time travel in sci-fi content or financial matters in documentary content. While not perhaps easily expressed as obvious and separate genres, these choices can be identified and taken into account in mathematical models.

It’s this contextual and behavioural awareness that can really make the difference and allow search to become a slick and engaging experience rather than a laborious one.

Light at the end of the tunnel

UX is already a key battleground for operators, and search and recommendation is but one front in bringing the latest experiences to customers. To bring next-generation user experience to their search services, operators need to look at other industries and how they are working to change the way that customers interact with them – for example search engines online and voice control systems on mobile devices.

By using these technologies, operators can provide a hyper-intelligent hyper-relevant and natural service to their customers.

Sashikumar Venkataraman, Director Engineering, TiVo (opens in new tab)

Image source: Shutterstock/isak55