Knowledge graphs enable a wide variety of applications, including question answering and information retrieval. Despite the great effort invested in their creation and maintenance, even the largest (e.g., Yago, DBPedia or Wikidata) remain incomplete. We introduce Relational Graph Convolutional Networks (R-GCNs) and apply them to two standard knowledge base completion tasks: Link prediction (recovery of missing facts, i.e. subject-predicate-object triples) and entity classification (recovery of missing entity attributes). R-GCNs are related to a recent class of neural networks operating on graphs, and are developed specifically to handle the highly multi-relational data characteristic of realistic knowledge bases. We demonstrate the effectiveness of R-GCNs as a stand-alone model for entity classification. We further show that factorization models for link prediction such as DistMult can be significantly improved through the use of an R-GCN encoder model to accumulate evidence over multiple inference steps in the graph, demonstrating a large improvement of 29.8% on FB15k-237 over a decoder-only baseline.

/pdf/modeling-relational-data-with-graph-convolutional-networks-81d86mzfgx.pdf

Modeling Relational Data with Graph Convolutional Networks

Knowledge graphs enable a wide variety of applications, including question answering and information retrieval. Despite the great effort invested in their creation and maintenance, even the largest (e.g., Yago, DBPedia or Wikidata) remain incomplete. We introduce Relational Graph Convolutional Networks (R-GCNs) and apply them to two standard knowledge base completion tasks: Link prediction (recovery of missing facts, i.e. subject-predicate-object triples) and entity classification (recovery of missing entity attributes). R-GCNs are related to a recent class of neural networks operating on graphs, and are developed specifically to deal with the highly multi-relational data characteristic of realistic knowledge bases. We demonstrate the effectiveness of R-GCNs as a stand-alone model for entity classification. We further show that factorization models for link prediction such as DistMult can be significantly improved by enriching them with an encoder model to accumulate evidence over multiple inference steps in the relational graph, demonstrating a large improvement of 29.8% on FB15k-237 over a decoder-only baseline.

Traditional dialog systems used in goal-oriented applications require a lot of domain-specific handcrafting, which hinders scaling up to new domains. End-to-end dialog systems, in which all components are trained from the dialogs themselves, escape this limitation. But the encouraging success recently obtained in chit-chat dialog may not carry over to goal-oriented settings. This paper proposes a testbed to break down the strengths and shortcomings of end-to-end dialog systems in goal-oriented applications. Set in the context of restaurant reservation, our tasks require manipulating sentences and symbols, so as to properly conduct conversations, issue API calls and use the outputs of such calls. We show that an end-to-end dialog system based on Memory Networks can reach promising, yet imperfect, performance and learn to perform non-trivial operations. We confirm those results by comparing our system to a hand-crafted slot-filling baseline on data from the second Dialog State Tracking Challenge (Henderson et al., 2014a). We show similar result patterns on data extracted from an online concierge service.

Learning End-to-End Goal-Oriented Dialog

Readings in information retrieval.

This paper formalizes the problem of solving multi-sentence algebraic word problems as that of generating and scoring equation trees. We use integer linear programming to generate equation trees and score their likelihood by learning local and global discriminative models. These models are trained on a small set of word problems and their answers, without any manual annotation, in order to choose the equation that best matches the problem text. We refer to the overall system as ALGES. We compare ALGES with previous work and show that it covers the full gamut of arithmetic operations whereas Hosseini et al. (2014) only handle addition and subtraction. In addition, ALGES overcomes the brittleness of the Kushman et al. (2014) approach on single-equation problems, yielding a 15% to 50% reduction in error.

/pdf/parsing-algebraic-word-problems-into-equations-3aarfakiy4.pdf

Parsing Algebraic Word Problems into Equations

We describe how a question-answering system can learn about its domain from conversational dialogs. Our system learns to relate concepts in science questions to propositions in a fact corpus, stores new concepts and relations in a knowledge graph (KG), and uses the graph to solve questions. We are the first to acquire knowledge for question-answering from open, natural language dialogs without a fixed ontology or domain model that predetermines what users can say. Our relation-based strategies complete more successful dialogs than a query expansion baseline, our taskdriven relations are more effective for solving science questions than relations from general knowledge sources, and our method is practical enough to generalize to other domains.

/pdf/learning-knowledge-graphs-for-question-answering-through-53uvw39z1q.pdf

Learning Knowledge Graphs for Question Answering through Conversational Dialog

As grapheme-to-phoneme methods proliferate, their careful evaluation becomes increasingly important. This paper explores a variety of metrics to compare the automatic pronunciation methods of three freely-available grapheme-to-phoneme packages on a large dictionary. Two metrics, presented here for the first time, rely upon a novel weighted phonemic substitution matrix constructed from substitution frequencies in a collection of trusted alternate pronunciations. These new metrics are sensitive to the degree of mutability among phonemes. An alignment tool uses this matrix to compare phoneme substitutions between pairs of pronunciations. Index Terms: grapheme-to-phoneme, edit distance, substitution matrix, phonetic distance measures

/pdf/phonemic-similarity-metrics-to-compare-pronunciation-methods-3zqxjsgh7g.pdf

Phonemic Similarity Metrics to Compare Pronunciation Methods

We present open dialogue management and its application to relational databases. An open dialogue manager generates dialogue states, actions, and default strategies from the semantics of its application domain. We define three open dialogue management tasks. First, vocabulary selection finds the intelligible attributes in each database table. Second, focus discovery selects candidate dialogue foci, tables that have the most potential to address basic user goals. Third, a focus agent is instantiated for each dialogue focus with a default dialogue strategy governed by efficiency. We demonstrate the portability of open dialogue management on three very different databases. Evaluation of our system with simulated users shows that users with realistically limited domain knowledge have dialogues nearly as efficient as those of users with complete domain knowledge.

/pdf/open-dialogue-management-for-relational-databases-2r2nbgxube.pdf

Open Dialogue Management for Relational Databases

Ambiguous or open-ended requests to a dialogue system result in more complex dialogues. We present a semantic-specificity metric to gauge this complexity for dialogue systems that access a relational database. An experiment where a simulated user makes requests to a dialogue system shows that semantic specificity correlates with dialogue length.

/pdf/semantic-specificity-in-spoken-dialogue-requests-ntb7pfcre5.pdf

Semantic Specificity in Spoken Dialogue Requests

Recommendations from a system to a person require thoughtful decisions. People prefer quick, relevant responses, ones that quickly lead to an acceptable choice. A system’s recommendations, however, often rely on pairwise similarity between items or between users, based on a distance metric. This paper advocates a broader, set-based similarity based on text descriptions of known items, and on the items users have found appealing in the past. The text, autonomously gathered from the Web, proves to be a strong indicator of similarities among sets of items, as well as a source of useful diversity. Empirical results provide insight into the complexities of similarity, of user behavior, and of cognition in the system that formulates recommendations.

/pdf/similarity-and-plausible-recommendations-3cn8fr7de7.pdf

Ben Hixon

Papers

Learning Knowledge Graphs for Question Answering through Conversational Dialog

Phonemic Similarity Metrics to Compare Pronunciation Methods

Open Dialogue Management for Relational Databases

Semantic Specificity in Spoken Dialogue Requests

Similarity and Plausible Recommendations