Relational retrieval using a combination of path-constrained random walks

@article{Lao2010RelationalRU,
  title={Relational retrieval using a combination of path-constrained random walks},
  author={N. Lao and William W. Cohen},
  journal={Machine Learning},
  year={2010},
  volume={81},
  pages={53-67}
}
Scientific literature with rich metadata can be represented as a labeled directed graph. This graph representation enables a number of scientific tasks such as ad hoc retrieval or named entity recognition (NER) to be formulated as typed proximity queries in the graph. One popular proximity measure is called Random Walk with Restart (RWR), and much work has been done on the supervised learning of RWR measures by associating each edge label with a parameter. In this paper, we describe a novel… 

Fast query execution for retrieval models based on path-constrained random walks

TLDR
The use of approximations to the computation of the PCRW distributions, including fingerprinting, particle filtering, and truncation strategies, are evaluated, showing speedups of factors of 2 to 100 with little loss in accuracy.

Learning Relation Networks for Relational Retrieval

TLDR
A thesis work on effectively learning complex retrieval strategies which can leverage rich structural information presented as heterogeneous graphs by generalizing the popularly used random walk with restart models so that the retrieval strategies are represented as relation paths or relation networks on the graph.

Query expansion using path-constrained random walks

TLDR
Results show that the PCRW-based method is very effective for the expansion of rare queries, i.e., low-frequency queries that are unseen in search logs, and that it outperforms significantly other state-of-the-art QE meth-ods.

Markov Logic Sets: Towards Lifted Information Retrieval Using PageRank and Label Propagation

TLDR
This work describes an algorithm that ranks atoms using a score based on random walks with restart (RWR): the probability that a random surfer hits an atom starting from the query atoms, and compares to Bayesian sets and shows that this approach gives very reasonable set completions.

Interactive Paths Embedding for Semantic Proximity Search on Heterogeneous Graphs

TLDR
This paper introduces a novel concept of interactive paths to model the inter-dependency among multiple paths between a query object and a target object, and proposes an Interactive Paths Embedding (IPE) model, which learns low-dimensional representations for the resulting interactive-paths structures for proximity estimation.

Personalized Navigation and Random Walk on a Complex Heterogeneous Graph

TLDR
While PGN can cope with information retrieval/recommendation problems at a low cost, this method also transforms a complex heterogeneous graph into a homogeneous graph, allowing existing homogeneousgraph mining algorithms to be applied to a heterogeneousgraph.

Semi-supervised Learning over Heterogeneous Information Networks by Ensemble of Meta-graph Guided Random Walks

TLDR
This paper decomposes the original HIN schema into several semantically meaningful meta-graphs consisting of entity and relation types and presents a semi-supervised learning algorithm constrained by the types of HINs.

Learning Relational Features with Backward Random Walks

TLDR
Cor-PRA is described, an enhanced system that can model a larger space of relational rules, including longer relational rules and a class of first order rules with constants with constants, while maintaining scalability.

Learning Relational Features with Backward Random Walks Submitted for Blind Review

  • Computer Science
  • 2012
TLDR
This paper proposes to compute backward random walks to efficiently model paths that correspond to first order rules with constants in this framework, and extends PRA to consider long relational paths.

Evaluating Top-k Meta Path Queries on Large Heterogeneous Information Networks

TLDR
A solution that seamlessly integrates several ranking functions that evaluate the importance of meta paths based on frequency and rarity, rather than on path length is proposed that outperforms state-of-the-art algorithms.
...

References

SHOWING 1-10 OF 30 REFERENCES

Learning Parameters in Entity Relationship Graphs from Ranking Preferences

TLDR
A unified model for ranking in ER graphs is presented, and an algorithm to learn the parameters of the model is proposed, which can satisfy training preferences and generalize to test preferences, and estimate meaningful model parameters that represent the relative importance of ER types.

Learning Graph Walk Based Similarity Measures for Parsed Text

TLDR
A new path-constrained graph walk method is proposed, in which the graph walk process is guided by high-level knowledge about meaningful edge sequences (paths), which shows that this framework is preferable to vector-based models for small-sized corpora.

ObjectRank: Authority-Based Keyword Search in Databases

Users, Queries and Documents: A Unified Representation for Web Mining

TLDR
This paper proposes a richer data structure which is able to preserve most of the information available in the logs including query refinements, page visits and search activity and shows that this more powerful representation can be successfully used to improve the quality of query clustering and to discover query suggestions.

Learning to rank networked entities

TLDR
A framework for ranking networked entities based on Markov walks with parameterized conductance values associated with the network edges and a constrained maximum entropy network flow formulation whose dual can be solved efficiently using a cutting-plane approach and a quasi-Newton optimizer.

Contextual search and name disambiguation in email using graphs

TLDR
This paper provides a detailed instantiation of this framework for email data, where content, social networks and a timeline are integrated in a structural graph and shows that reranking schemes based on the graph-walk similarity measures often outperform baseline methods and that further improvements can be obtained by use of appropriate learning methods.

Object-level ranking: bringing order to Web objects

TLDR
The experimental results show that PopRank can achieve significantly better ranking results than naively applying PageRank on the object graph, and the proposed efficient approaches to automatically decide these factors are proposed.

Learning Web Page Scores by Error Back-Propagation

TLDR
A novel algorithm to learn a score distribution over the nodes of a labeled graph (directed or undirected) and the effectiveness of the proposed technique in reorganizing the rank accordingly to the examples provided in the training set is shown.

Learning random walk models for inducing word dependency distributions

TLDR
A Markov chain model, whose stationary distribution is used to give word probability estimates, is proposed, which is applied to the task of prepositional phrase attachment, obtaining an accuracy of 87.54%.

Document selection methodologies for efficient and effective learning-to-rank

TLDR
This paper employs a number of document selection methodologies, widely used in the context of evaluation--depth-k pooling, sampling, sampling (infAP, statAP), active-learning (MTC), and on-line heuristics (hedge), and investigates whether they can enable efficient and effective learning-to-rank.