StatsMonkey: A Data-Driven Sports Narrative Writer
StatsMonkey: A Data-Driven Sports Narrative Writer
复制标题
StatsMonkey：数据驱动的体育叙事作家
DOI：
10.4230/oasics.cmn.2013.106
复制
发表时间：
2010
期刊：
影响因子：
--
通讯作者：
K. Hammond
中科院分区：
文献类型：
--
作者：
Nicholas D. Allen;John R. Templon;P. McNally;L. Birnbaum;K. Hammond
有一些特定类型的故事通常是以非常结构化的方式讲述的；体育新闻或财务报道就是两个例子。读者关心这些叙述，因为他们对这个话题非常感兴趣，想要了解事件的具体细节。换句话说，他们关心数据，并希望阅读向他们展示这些数据的故事。然而，为了引人注目，这些叙述不能只是重复数据，而是必须从数据中讲述一个故事。在本文中，我们将介绍一个数据驱动的故事讲述模型，并讨论StatsMonkey，这是一个根据在线可用的原始棒球比赛数字数据自动编写棒球故事的系统。我们将展示一台机器可以生成有趣的、可读的故事，并且它可以做出编辑决定，决定突出显示情况的哪些方面。进一步，我们将展示一台机器可以决定以何种方式共享这些方面。
There are certain types of stories that are often told in very structured ways; sports stories or financial reports are two examples. Readers care about these narratives because they are passionately interested in the topic and want to read about the specific details of the event. In other words, they care about the data and want to read a story that presents that data to them. However, in order to be compelling these narratives cannot merely repeat the data, rather they must tell a story from the data. In this paper, we will present a model for data-driven story-telling and discuss StatsMonkey, a system that automatically writes baseball stories from raw baseball game numerical data available online. We will show that a machine can generate interesting, readable stories and that it can make editorial decisions about what aspects of a situation to highlight. Further we will show that a machine can determine in what manner those aspects should be shared.