Paper: arXiv 2407.10909

Abstract

Dynamic knowledge graphs (DKGs) are popular structures to express different types of connections between objects over time. They can also serve as an efficient mathematical tool to represent information extracted from complex unstructured data sources, such as text or images. Within financial applications, DKGs could be used to detect trends for strategic thematic investing, based on information obtained from financial news articles. In this work, we explore the properties of large language models (LLMs) as dynamic knowledge graph generators, proposing a novel open-source fine-tuned LLM for this purpose, called the Integrated Contextual Knowledge Graph Generator (ICKG). We use ICKG to produce a novel open-source DKG from a corpus of financial news articles, called FinDKG, and we propose an attention-based GNN architecture for analysing it, called KGTransformer. We test the performance of the proposed model on benchmark datasets and FinDKG, demonstrating superior performance on link prediction tasks. Additionally, we evaluate the performance of the KGTransformer on FinDKG for thematic investing, showing it can outperform existing thematic ETFs.

Complexity vs Empirical Score

  • Math Complexity: 7.5/10
  • Empirical Rigor: 6.0/10 (capped from 7.0: transaction costs not modelled; see penalty flags)
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper introduces novel mathematical architectures (KGTransformer based on GATs/HGTs/EvoKG) with formal definitions of dynamic knowledge graphs, while also providing concrete empirical evidence including a fine-tuned LLM pipeline, open-source datasets (FinDKG), and performance metrics on link prediction and thematic investing vs. ETFs.

Research Flowchart

  flowchart TD
  A["Research Goal:<br/>Detect Global Financial Trends<br/>using Dynamic Knowledge Graphs"] --> B["Data: Financial News Corpus"]
  B --> C{"Core Methodology:<br/>Generate & Analyze DKG"}
  C --> D["LLM Generator<br/>(ICKG):<br/>Build Dynamic Knowledge Graph"]
  C --> E["Graph Neural Network<br/>(KGTransformer):<br/>Attention-based Analysis"]
  D --> E
  E --> F["Computational Task:<br/>Link Prediction<br/>& Thematic Investing"]
  F --> G["Key Findings:<br/>1. SOTA Link Prediction<br/>2. Outperforms Thematic ETFs<br/>3. Validated FinDKG"]

Cited References (145)

  1. Heterogeneous Graph Transformer (2020) — link
  2. FinGPT: Open-Source Financial Large Language Models — → 2306.06031
  3. LLaMA: Open and Efficient Foundation Language Models (2023) — link
  4. Graph Attention Networks (2018) — link
  5. Decoupled Weight Decay Regularization (2019)
  6. Mistral 7B (2023) — link
  7. The patent holder’s dilemma: Buy, sell, or troll? (2007) — link
  8. Deciding equivalances among conjunctive aggregate queries (2007) — link
and 137 more
  1. Special issue: Digital Libraries (1996)
  2. Understanding Policy-Based Networking (2001)
  3. First-Order Dynamic Logic (1979) — link
  4. The title of book one (2007) — link
  5. The title of book two (2008) — link
  6. Achieving application requirements (1990) — link
  7. Statecarts in use: structured analysis and object-orientation (1998) — link
  8. The Art of Computer Programming, Vol. 1: Fundamental Algorithms (3rd. ed.) (1997)
  9. The Art of Computer Programming (1998)
  10. An experiment in bibliographic mark-up: Parsing metadata for XML export (2010) — link
  11. Catch me, if you can: Evading network signatures with web-based polymorphic worms (2007)
  12. Catch me, if you can: Evading network signatures with web-based polymorphic worms (2008)
  13. Catch me, if you can: Evading network signatures with web-based polymorphic worms (2009)
  14. Predicate Path expressions (1979) — link
  15. LOGICS of Programs: AXIOMATICS and DESCRIPTIVE POWER (1978)
  16. Optimal Motion Control of a Ground Vehicle (2003)
  17. Algorithms for Closest-Point Problems (Computational Geometry) (1985)
  18. Introduction to Bayesian Statistics (2001) — link
  19. CLIFFORD: a Maple 11 Package for Clifford Algebra Computations, version 11 (2007) — link
  20. Stats and Analysis (2006) — link
  21. A more perfect union (2008) — link
  22. The fountain of youth (2009)
  23. Solder man (2003) — link
  24. Interview with Bill Kinder: January 13, 2005 (2005) — link
  25. The Enabling of Digital Libraries (2008)
  26. (old) Finding minimum congestion spanning trees (2000) — link
  27. (new) Finding minimum congestion spanning trees (2000) — link
  28. (old) Distributed data source verification in wireless sensor networks (2009) — link
  29. (new) Distributed data source verification in wireless sensor networks (2009) — link
  30. Portalis: using competitive online interactions to support aid initiatives for the homeless (2008) — link
  31. Visual Basic 6: Design, Specification, and Objects with Other (1999)
  32. The Latex Web Companion: Integrating TEX, HTML, and XML (1999)
  33. Vertex Types in Book-Embeddings (1987)
  34. Vertex Types in Book-Embeddings (1987)
  35. CHI ‘08: CHI ‘08 extended abstracts on Human factors in computing systems (2008)
  36. Algorithms for Closest-Point Problems (Computational Geometry) (1985)
  37. IEEE TCSC Executive Committee (2004) — link
  38. Distributed systems (2nd Ed.) (1993)
  39. New Algorithms for Dependency-Directed Backtracking (Master’s thesis) (1986)
  40. New Algorithms for Dependency-Directed Backtracking (Master’s thesis) (1986)
  41. Seminumerical Algorithms (1981)
  42. E-commerce and cultural values (2001) — link
  43. Chapter 9 (2002) — link
  44. The implementation of electronic commerce in SMEs in Singapore (Incoll) (2003) — link
  45. E-commerce and cultural values - (InBook-num-in-chap) (2004) — link
  46. E-commerce and cultural values (Inbook-text-in-chap) (2005) — link
  47. E-commerce and cultural values (Inbook-num chap) (2006) — link
  48. A library-based synthesis methodology for reversible logic (2010)
  49. Synthesis of Reversible Circuit Using Cycle-Based Approach (2010)
  50. Algorithmic Enumeration of Ideal Classes for Quaternion Orders (2010) — link
  51. Chapter II: Notes on data structuring (1972) — link
  52. Transcript of question and answer session (1981) — link
  53. Go to statement considered harmful (1979) — link
  54. Three-dimensional virtual acoustic displays (1992) — link
  55. Managerial expert systems and organizational change: some critical research issues (1987) — link
  56. Simplified Structured COBOL with Microsoft/MicroFocus COBOL (1990)
  57. A multi-radio unification protocol for IEEE 802.11 wireless networks (2004)
  58. Wireless Sensor Networks: A Survey (2002)
  59. A Survey on Wireless Multimedia Sensor Networks (2007)
  60. SSCH : Slotted Seeded Channel Hopping for Capacity Improvement in IEEE 802.11 Ad-Hoc Wireless Networks (2004)
  61. XBOW Sensor Motes Specifications (2008)
  62. Overview of Sensor Networks (2004)
  63. CodeBlue : Sensor Networks for Medical Care (2008)
  64. Investigating Network Architectures for Body Sensor Networks (2007) — link
  65. Channel-Hopping Multiple Access (2000)
  66. Body Sensor Networks (2008)
  67. Mapping Powerlists onto Hypercubes (1994)
  68. Automatic Parallelization for Distributed-Memory Multiprocessing Systems (1989)
  69. User recovery and reversal in interactive systems (1984)
  70. Generalizing specifications for uniformly implemented loops (1985)
  71. Towards monolingual programming environments (1985)
  72. The \TeX book (1984)
  73. Distributed algorithms for finding centers and medians in networks (1984)
  74. \it \LaTeX : A Document Preparation System (1986)
  75. Program transformations in a denotational setting (1985)
  76. Seminumerical Algorithms (1981)
  77. A high-level approach to computer document formatting (1980)
  78. A multifrequency MAC specially designed for wireless sensor network applications (2010) — link
  79. Institutional members of the \TeX Users Group (2017) — link
  80. acmart— C lass for typesetting publications of ACM (2017) — link
  81. Reasoning About Naming Systems (1993) — link
  82. Babel, a Multilingual Style-Option System for Use with LaTeX’s Standard Document Styles (1991)
  83. A Methodology for Implementing Highly Concurrent Data Objects (1993) — link
  84. Publication quality tables in \LaTeX (2005)
  85. Using the amsthm Package (2015)
  86. R: A Language and Environment for Statistical Computing (2019) — link
  87. UMass Citation Field Extraction Dataset (2013) — link
  88. Citation concept analysis (CCA) —A new form of citation analysis revealing the usefulness of concepts for other researchers illustrated by two exemplary case studies including classic books by Thomas S.~Kuhn and Karl R.~Popper (2019) — link
  89. Learning Soft Linear Constraints with Application to Citation Field Extraction (2014) — link
  90. Maintaining Discrete Probability Distributions Optimally (1993)
  91. Networks and the macroeconomy: An empirical exploration (2016)
  92. FinBERT: Financial sentiment analysis with pre-trained language models (2019) — link
  93. Measuring economic policy uncertainty (2016)
  94. Large-scale simple question answering with memory networks (2015) — link
  95. ICEWS Coded Event Data (2015) — link
  96. A survey on temporal knowledge graph: representation learning and applications (2024) — link
  97. Graph representation learning: a survey (2020)
  98. Expected returns and large language models (2022)
  99. Knowledge graph-based event embedding framework for financial quantitative investments (2020)
  100. Modeling momentum spillover with economic links discovered from financial documents (2023)
  101. Temporal relational ranking for stock prediction (2019)
  102. Stochastic optimization for market return prediction using financial knowledge graph (2018)
  103. Learning sequence encoders for temporal knowledge graph completion (2018) — link
  104. Text as data (2019)
  105. Large language model adaptation for financial sentiment analysis (2024) — link
  106. A survey on knowledge graphs: representation, acquisition, and applications (2021)
  107. Recurrent event network: autoregressive structure inference over temporal knowledge graphs (2019) — link
  108. A survey on graph representation learning methods (2024)
  109. Semi-supervised classification with graph convolutional networks (2016) — link
  110. Deriving validity time in knowledge graph (2018)
  111. Gdelt: Global data on events, location, and tone, 1979–2012 (2013)
  112. Can chatgpt forecast stock price movements? return predictability and large language models — → 2304.07619
  113. Textual analysis in accounting and finance: A survey (2016)
  114. A Survey of Large Language Models for Financial Applications: Progress, Prospects and Challenges — → 2406.11903
  115. GPT-4 technical report (2023)
  116. Modal-adaptive Knowledge-enhanced Graph-based Financial Prediction from Monetary Policy Conference Calls with LLM (2024) — link
  117. EvoKG: Jointly modeling event time and network structure for reasoning over temporal knowledge graphs (2022)
  118. Sentence-bert: Sentence embeddings using siamese bert-networks (2019) — link
  119. Knowledge graphs: an information retrieval perspective (2020)
  120. HybridRAG: Integrating Knowledge Graphs and Vector Retrieval Augmented Generation for Efficient Information Extraction — → 2408.04948
  121. Modeling relational data with graph convolutional networks (2018)
  122. Yago: a core of semantic knowledge (2007)
  123. Know-evolve: Deep temporal reasoning for dynamic knowledge graphs (2017)
  124. Graph denoising networks: a deep learning framework for equity portfolio construction (2023)
  125. Attention is all you need (2017)
  126. Graph attention networks (2017) — link
  127. RippleNet: Propagating user preferences on the knowledge graph for recommender systems (2018)
  128. Knowledge-aware graph neural networks with label smoothness regularization for recommender systems (2019)
  129. Anti-money laundering in Bitcoin: experimenting with graph convolutional networks for financial forensics (2019) — link
  130. An efficient gradient-based algorithm for on-line training of recurrent network trajectories (1990)
  131. How powerful are graph neural networks? (2018) — link
  132. The knowledge graph for macroeconomic analysis with alternative big data (2020) — link
  133. Pre-trained Embeddings for Entity Resolution: An Experimental Analysis (2023) — link
  134. Instruct-FinGPT: Financial sentiment analysis by instruction tuning of general-purpose large language models — → 2306.12659
  135. ICEWS Coded Event Data (2015) — link
  136. Yago: a core of semantic knowledge (2007)
  137. Wikidata: a free collaborative knowledgebase (2014) — link

5 of 145 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2518 linked papers.