Virtual Book Browsing and AI-powered Book Recommendations

“虚拟图书浏览和人工智能图书发现”项目将为英国书店和书商提供创新的工具和服务，如深入查看、稍后阅读和白标人工智能图书推荐服务，以增强他们在混乱后的在线和电子商务能力。该项目是出版技术专家Jellybook和英国第二大实体图书连锁店Blackwell‘s的合作伙伴，后者也将是该项目的试点合作伙伴。该项目的核心是一个全国性的电子书样本基础设施，向任何英国书店或书商开放。该平台将支持在线浏览一本书的前5%到20%(Look-Inside)，包括离线阅读支持(稍后阅读)。每个展示的图书样本都将有一个集成的购买链接，为用户提供从图书中购买数字或纸质版本(点击购买)的选项，包括店内提货(点击并收集)的选项。平台将使用Jellybook零售商本地化引擎作为闭环系统运行。这意味着，与其他解决方案不同，零售商A提供和分发的电子书样本只有相同零售商A的购买链接，而不是竞争对手零售商B和C的购买链接，即使B和C可能也将该平台用于自己的目的，甚至销售相同的图书。消费者将不需要下载电子书文件或安装应用程序。取而代之的是，该平台将利用最先进的Jellybook Cloud Reader，在任何支持浏览器的连接设备上支持云阅读。Jellybook Cloud Reader基于开放的ePub标准，能够支持视障读者和其他有无障碍要求的读者。通过国家平台从零售商那里收集的数据将用于开发和训练机器学习算法(俗称人工智能或AI)。其目标是识别品味集群，从而更好、更准确地描述图书的类别。使用观察性用户数据将纠正现有分类系统中固有的行业偏见和错误，支持更多样化的作者和声音，并减少当前以大众市场为重点的系统中可能发生的受众和作者的边缘化。机器学习算法还将被用于为图书零售商和出版商创建图书推荐引擎。这将采取白标直接面向消费者的时事通讯服务的形式，将生成针对个人用户、社区和受众利基的人工智能支持的图书推荐。简而言之，该服务将提供Netflix式的推荐引擎，以适应和优化独立图书零售商的分散需求，这是一项重大创新。该平台还将用于推动用户选择产生最少包装浪费和最小碳足迹的购买路径。该项目将于2021年初向最终用户开放。

The project "Virtual Book Browsing and AI-powered Book Discovery" will provide UK bookshops and booksellers with innovative tools and services, such as look-inside, read later and a white-label AI-powered book recommendation service to enhance their online and ecommerce capabilities post pandemic.The project is a partnership between publishing technology specialist Jellybooks and bookseller Blackwell's, the UK's second largest physical book chain, who will also be the pilot partner in this project.At the core of the project is a national ebook sample infrastructure open to any UK bookshop or bookseller. This platform will support online browsing of the first 5% to 20% of a book (look-inside) including offline reading support (read later). Every book sample featured will have an integrated buy link offering users the option from within the book to purchase a digital or physical version (click-to-buy), including the option of in-store pick-up (click & collect).The platform will use the Jellybooks retailer localisation engine to operate as a closed-loop system. This means that, unlike alternative solutions, an ebook sample offered and distributed by retailer A only has buy links to the same retailer A and not competing retailers B and C even though B and C may be using the platform for their own purposes as well, even selling the same books.Consumers will not be required to download an ebook file or install an app. Instead the platform will leverage the state-of-the-art Jellybooks Cloud Reader to support reading in the cloud on any connected device with browser support. The Jellybooks Cloud Reader is based on the open ePub standard and is able to support visually impaired readers and other readers with accessibility requirements.Data collected across retailers via the national platform will be used to develop and train machine learning algorithms (colloquially referred to as artificial intelligence or AI). The goal is to identify taste clusters allowing for better and more accurate category description of books. The use of observational user data will correct for inherent industry biases and errors in existing classification systems, support a greater diversity of authors and voices, and reduce the marginalisation of audiences and authors that can occur in the current mass-market focused system.The machine learning algorithms will also be deployed to create a book recommendation engine for book retailers and publishers. This will take the shape of a white-label direct-to-consumer newsletter service that will generate AI-powered book recommendations specific to individual users, communities, and audience niches. In short, the service will offer a Netflix-style recommendation engine adapted and optimised for the decentralised needs of independent book retailers, a major innovation.The platform will also be used to nudge users towards the purchase path generating the least packaging waste and the smallest carbon footprint in terms of book delivery.The project pilot will open to end users at the beginning of 2021\.