Algorithms for storytelling

Abstract

We formulate a new data mining problem called storytelling as a generalization of redescription mining. In traditional redescription mining, we are given a set of objects and a collection of subsets defined over these objects. The goal is to view the set system as a vocabulary and identify two expressions in this vocabulary that induce the same set of objects. Storytelling, on the other hand, aims to explicitly relate object sets that are disjoint (and hence, maximally dissimilar) by finding a chain of (approximate) redescriptions between the sets. This problem finds applications in bioinformatics, for instance, where the biologist is trying to relate a set of genes expressed in one experiment to another set, implicated in a different pathway. We outline an efficient storytelling implementation that embeds the CARTwheels redescription mining algorithm in an A* search procedure, using the former to supply next move operators on search branches to the latter. This approach is practical and effective for mining large datasets and, at the same time, exploits the structure of partitions imposed by the given vocabulary. Three application case studies are presented: a study of word overlaps in large English dictionaries, exploring connections between genesets in a bioinformatics dataset, and relating publications in the PubMed index of abstracts.

Interactive Discovery of Coordinated Relationship Chains with Maximum Entropy Models

“…Most importantly, we consider data matrices where a row (or, document) may be so sparse or small (e.g., 1-paragraph snippets) that it is difficult to calculate statistically meaningful scores. Storytelling algorithms [e.g., 19,29,20] are another related thread of research; they provide algorithmic ways to rank connections between entities but do not focus on entity coalitions and how such coalitions are maintained through multiple sources of evidence. Wu et al [57] proposed a framework to discover the plots by detecting non-obvious coalitions of entities from multi-relational datasets with Maximum Entropy principle and further support iterative, human-in-the-loop, knowledge discovery.…”

SGSG: Semantic graph-based storyline generation in Twitter

“…Although, the problem of storytelling has recently been focused on in Twitter, it was first formulated in Kumar et al [17] as a generalisation of redescription mining. Given a set of objects and a collection of subsets over these objects, they relate the object sets that are disjoint and dissimilar by finding a chain of redescriptions between the sets.…”

Analyzing evolving stories in news articles

“…e main component of our framework is focused on the problem of storytelling, or connecting the dots. ere are several approaches that aim to solve this problem for di erent types of datasets, such as scienti c articles [15], entity networks [7,8,13], image collections [12,32] and document collections [1,10,11,20,24,30,31,34,40]. Most of the work on this problem uses graph-based representations of a document or entity set [7,10,14,24,40].…”

“…While capturing such di usions over a timeline is still a challenge, researchers have targeted the problem of tracking stories in di erent guises, e.g., storytelling [14,20], storyboarding [23], connecting the dots [15,30], and metro maps [31]. Most of these methods nd underlying connections between articles using a similaritybased network of documents.…”