Concept Expansion Using Web Tables

Abstract

We study the following problem: given the name of an ad-hoc concept as well as a few seed entities belonging to the concept, output all entities belonging to it. Since producing the exact set of entities is hard, we focus on returning a ranked list of entities. Previous approaches either use seed entities as the only input, or inherently require negative examples. They suffer from input ambiguity and semantic drift, or are not viable options for ad-hoc tail concepts. In this paper, we propose to leverage the millions of tables on the web for this problem. The core technical challenge is to identify the "exclusive" tables for a concept to prevent semantic drift; existing holistic ranking techniques like personalized PageRank are inadequate for this purpose. We develop novel probabilistic ranking methods that can model a new type of table-entity relationship. Experiments with real-life concepts show that our proposed solution is significantly more effective than applying state-of-the-art set expansion or holistic ranking techniques.

SmartTable

“…It is closely related to the task of entity set expansion [2,5,7], which is about expanding a seed entity set with additional instances. The main difference between row population and entity entity set expansion is that we can also leverage additional data from the seed table as input, not only the core column entities.…”

EntiTables

“…He and Xin [10] focus on entity list data, by picking the top-k entities based on relevance between the candidate entity and seed entities, and then iteratively ranking them according to a combination of relevance and coherence. Similar iterative steps are conducted in [26,29]. Wang and Cohen [29] use a random walk method for ranking during iterations.…”

“…One line of prior work [4,16] relies on a knowledge base for establishing this semantic similarity. Another family of approaches [8,26,29] leverages a large table corpus for collecting co-occurrence statistics. We combine both these sources in a single model:…”

“…We also introduce a number of innovative elements on the component level. e task of row population relates to the task of entity set expansion [4,8,10,16,26,29], where a given set of seed entities (examples) is to be completed with additional entities. We also have a seed set of entities from the le most table column.…”

SetExpan: Corpus-Based Set Expansion via Context Feature Selection and Rank Ensemble

“…While this approach can achieve relatively good quality, the required seed-oriented online data extraction is costly. Therefore, more studies [17] [23][10] [28] [21] are proposed in a corpus-based setting where sets are expanded by offline processing based on a specific corpus. For corpus-based set expansion, there are two general approaches, one-time entity ranking and iterative pattern-based bootstrapping.…”