Timely Semantics: A Study of a Stream-Based Ranking System for Entity Relationships

Abstract

In recent years, search engines have started presenting se-mantically relevant entity information together with document search results. Entity ranking systems are used to compute recommendations for related entities that a user might also be interested to explore. Typically, this is done by ranking relationships between entities in a semantic knowledge graph using signals found in a data source as well as type annotations on the nodes and links of the graph. However, the process of producing these rankings can take a substantial amount of time. As a result, entity ranking systems typically lag behind real-world events and present relevant entities with outdated relationships to the search term or even outdated entities that should be replaced with more recent relations or entities. This paper presents a study using a real-world stream-processing based implementation of an entity ranking system, to understand the effect of data timeliness on entity rankings. We describe the system and the data it processes in detail. Using a longitudinal case-study, we demonstrate (i) that low-latency, large-scale entity relationship ranking is feasible using moderate resources and (ii) that stream-based entity ranking improves the freshness of related entities while maintaining relevance. Abstract. In recent years, search engines have started presenting semantically relevant entity information together with document search results. Entity ranking systems are used to compute recommendations for related entities that a user might also be interested to explore. Typically, this is done by ranking relationships between entities in a semantic knowledge graph using signals found in a data source as well as type annotations on the nodes and links of the graph. However, the process of producing these rankings can take a substantial amount of time. As a result, entity ranking systems typically lag behind real-world events and present relevant entities with outdated relationships to the search term or even outdated entities that should be replaced with more recent relations or entities. This paper presents a study using a real-world stream-processing based implementation of an entity ranking system, to understand the effect of data timeliness on entity rankings. We describe the system and the data it processes in detail. Using a longitudinal case-study, we demonstrate (i) that low-latency, large-scale entity relationship ranking is feasible using moderate resources and (ii) that stream-based entity ranking improves the freshness of related entities while maintaining relevance.

A Comprehensive Survey of Knowledge Graph-Based Recommender Systems: Technologies, Development, and Contributions

“…• Dynamic data. Fischer et al [72] proposed a system that recommends entities by taking advantage of the temporal nature of search log data. This approach can significantly improve the quality of recommendations compared to certain static models of relevance; particularly, it improves the freshness measure.…”

“…Reusing of existing technologies and models. The architecture of [72] is based on previous work. On the other hand, Reference [19] proposed pre-trained or built models that can be run without training, thereby allowing faster deployment in new domains.…”

Multiple Models for Recommending Temporal Aspects of Entities

“…They extract several features from a variety of data sources and use a machine learning model to recommend entities to a Web search query. Following Spark, Sundog [9] aims to improve entity recommendation, in particular with respect to freshness, by exploiting Web search log data. The system uses a stream processing based implementation.…”

“…Beyond the traditional "ten blue links", to enhance user experience with entityaware intents, search engines have started including more semantic information, (1) suggesting related entities [4,9,30,31], or (2) supporting entity-oriented query completion or complex search with additional information or aspects [1,22,26]. These aspects cover a wide range of issues and include (but are not limited to) types, attributes/properties, relationships or other entities in general.…”

Machines Tuning Machines: Configuring Distributed Stream Processors with Bayesian Optimization

“…The first topology is a modified version of the Sundog entity ranking system [23]. Entity ranking systems consume search logs, tweets, etc., and rank search results based on cooccurence statistics.…”