AI in the publishing industry creates the opportunity to take published content to a higher level. AI can help with editorial processes, manage editorial workflows, and improve promotion strategies. In essence, it can massively help people without replacing them.
Due to the growing amount of digital information and ways of its representation, many publishers face an issue of reaching to their audience. Technology can help with this challenge enhancing content creation, distribution, and consumption.
Today, information is valuable as long as it is delivered quickly
Most major publishers have switched to digital and implemented machine learning algorithms to ensure that the end user gets tailored content in the shortest possible time.
Industry challenges are more and more often being solved with AI-powered tools. They are crucial for the future of journalism.
AI solutions for publishing industry
AI-based tools are already known to simplify processes such as grammar checking, editing, and formatting. Also, they can increase the accessibility of content in different languages – thanks to AI-based translation automation tools. What’s more, using natural language processing (NLP), we can reshape longer texts into shorter and more cohesive ones. Together with accelerating content creation, AI tools bring advantage of eliminating the risk of human error.
AI solutions in leading publishing companies
The implementation of AI-based tools by the most significant publishers are becoming the order of the day. Here are some examples:
Forbes introduced an AI-based CMS (content management system) called Bertie. It is an artificially intelligent publishing platform designed specifically for their internal newsroom consisted from journalists, experts, and partners.
The system provides them with popular real-time topics to discuss, recommending ways to make headlines more attractive and suggesting relevant images. Forbes announced the regular implementation of new AI technologies to facilitate the work of its employees.
Washington Post launched the Heliograf, an in-house technology built for hyperlocal service. Heliograf automates news writing, supporting journalists and reporters in their work. They started using it in data-grounded areas like sports and finance.
Heliograf was tested during the Olympics in 2016. News stories were put together by analysing games data. Next, Heliograf was matching the data to relevant phrases in a story to develop content for different platforms. The system wrote 850 articles in a year, which generated more than 500.000 online views. Continuous improvements have enabled to automatically write articles according to the Washington Post’s editorial line.
Bloomberg uses a solution called Cyborg. The system helps in creating and managing content. It supports journalists and editors mainly in the preparation of financial information. For example, it can generate thousands of articles about the company profit reports at the end of each quarter. Cyborg is programmed to identify and extract all keywords instantly.
5 AI use cases in publishing industry
1. Trend Driven Journalism
To maintain a competitive advantage in the publishing market, journalists and editors should provide interesting and reliable content as quickly as possible. News editors and journalists, who write about current events, usually do not have a problem with finding current, catchy topics. But the editors, who are specialists in specific fields, struggle with this challenge on a daily basis.
Valuable content is a sum of the publisher’s profile, data collecting speed, and audience-tailored topics
Manual searching for credible and valuable content is very time-consuming. Live reports, instant tweets, and thousands of sources of information of varying quality and credibility can’t be analysed in a reasonable time. In addition, only some news fit the publishing company’s profile, recipients’ interests, and advertisers’ expectations.
That’s why we created Trend Driven Journalism system. It is a tool supporting journalists and editors in finding right topics for their texts. TDJ is based on Google Trends information, which records what phrases people search for every day. Writers do not have to leave their system to search for information because they have Trends built into the recommendation tool.
Trend Driven Journalism’s features for publishing:
- Prompting the right topics based on publisher’s internal database, related platforms, and news portals.
- Recommending content ideas based on its popularity on the Internet.
- Help in choosing perfectly matched keywords and gaining valuable backlinks.
- Building strong content with correct quotations and semantic connections between published articles.
- Support in building strong visibility in search engines and generating high CTR.
2. SEO Tool
To improve SEO processes for publishers, AI can take over some activities. It can research keywords and optimise their use in published texts, and as a result, significantly increase the relevance and visibility of content. In effect, it can bring:
- high position in search engines;
- a large number of visits to the website;
- increased click-through rate (CTR) of advertisements.
However, keyword relevance is known to change rapidly, and each publisher’s news department has its own characteristics and significant subsets of keywords.
To remedy this, we created SEO Tool dedicated primarily to publishing companies. It is based on the so-called crawlers, which are modules responsible for constant analysing various sources of information in relation to their position in search engine results.
SEO Tool’s features for publishing:
- Support in the process of identifying and monitoring keywords – the tool shows a set of helpful metrics and divides keywords into different areas.
- Possibility to both independently determine monitored keywords and use automatically recommended phrases based on their popularity throughout the internet.
3. Deep Content Understanding
DCU is a tool that supports journalists in creating articles or editors in redaction. The tool suggests what and how to use it to enrich the content by generating and proposing connections between topics.
Our tool contains a knowledge base based on the so-called named-entities. They can be a time, a surname, a name of the city or an organisation. A combination of them determines what the text is about. This way, a network of links is formed between named-entity, articles already written, and even photos and people.
It finds articles using neural network models that detect the meaning of individual phrases throughout the text. The more extensive the database of content is, the more hints named-entity can generate. In addition, they can be narrowed down to a given period.
4. Content Moderation Platform
It is a tool that uses natural language processing to classify a given statement (phrase, comment) correctly. Then, thanks to in-depth analysis, it determines whether a given comment can be published on the website.
Our tests show that the system makes mistakes less often than people working as moderators. Probably it is because people manually reviewing and classifying several hundred comments a day lose their attention and vigilance after a few hours of work. Now, these people can take on the role of animators, enlivening discussion and building community.
Content Moderation Platform’s features for publishing:
- Helping to fight hate speech on the web.
- Rejecting comments saturated with negative emotions (the tool detects intentions even if a comment does not contain vulgarisms).
- Relieving moderators who manually approve or reject comments in favour of automatic moderation (now, they can become discussion animators).
- Blocking traffic coming from robots.
5. Article recommendations
Due to the amount of information that can be served to people, the work of online publishers is increasingly supported by recommendation systems. These systems aim to improve users’ access to materials tailored to their interests.
Recommendation systems can use AI to recommend articles and other resources to their readers based on the analysis of their historical preferences, interaction on the platform, or material information. Such personalisation improves user satisfaction, resulting in more significant interaction with the platform.
Recommendation systems are usually based on the following approaches:
- Content-based – the recommender analyses the similarity between content. Based on information about a given material (in the case of an article: title, content, author, category, etc.), it recommends similar content to the one that the given user liked or read.
- Collaborative-filtering – the recommender analyses the similarity between user, but also click-through rates and users preferences towards the materials. Based on similarities between users it recommends the same materials to people who show similar characteristics.
The newest tools aggregate these two approaches, so the results can be further improved.
Nowadays, without the help of AI it’s close to impossible to stay competitive in the publishing industry. By supporting editorial work and helping people in the tedious work of browsing hundreds of websites and sources, publishers increase their competitive advantage by using AI tools. Thanks to them, they can create content tailored to their readers and in line with current trends.
WEBSENSA offers all the above tools. Their uniqueness is based on an innovative combination of artificial intelligence, graph databases, and tracking trends on the web. We designed them on the basis of cloud computing, providing a scalability and ease of integration with the publisher’s systems – unrivalled at the moment.
AI tools for publishing industry – contact
If you are thinking about implementing any of the above solutions on your publishing platform and are looking for an experienced technology supplier, please contact: WEBSENSA – Contact. Our experts will carefully listen to your needs and help define the right solution.