公开(公告)号：US20250112878A1

公开(公告)日：2025-04-03

申请号：US18478702

申请日：2023-09-29

Applicant: Amazon Technologies, Inc.

Inventor： Samuel Bayless ,  Nadia Labai ,  Ora Yrjo Lassila

IPC: H04L51/02 ,  G06F40/20

Abstract: Techniques for a knowledge-graph system to use large language models (LLMs) to build knowledge graphs to answer queries submitted to a chatbot by users. The knowledge-graph system builds the knowledge graph using answers produced by an LLM for novel queries. The chatbot will continue to use the LLM to answer novel queries, but the chatbot may harness the knowledge graph to answer repeat questions to gain various efficiencies over LLM-backed chatbots. For example, the knowledge-graph system may easily debug or otherwise improve the answers in knowledge graphs, store provenance information in knowledge graphs, and augment the knowledge graphs using other data sources. Thus, the reliability and correctness of chatbots will be improved as the bugs and inaccuracies in answers provided by the LLM will be corrected in the knowledge graphs, but the chatbots can still harness the abilities of LLMs to provide answers across various subject-matter domains.

公开(公告)号：US20250111192A1

公开(公告)日：2025-04-03

申请号：US18375256

申请日：2023-09-29

Applicant: Amazon Technologies, Inc.

Inventor： Samuel Bayless ,  Nadia Labai ,  Ora Yrjo Lassila

IPC: G06N3/042 ,  G06N3/006 ,  G06N3/0455

Abstract: Techniques for a knowledge-graph system to use large language models (LLMs) to build knowledge graphs to answer queries submitted to a chatbot by users. The knowledge-graph system builds the knowledge graph using answers produced by an LLM for novel queries. The chatbot will continue to use the LLM to answer novel queries, but the chatbot may harness the knowledge graph to answer repeat questions to gain various efficiencies over LLM-backed chatbots. For example, the knowledge-graph system may easily debug or otherwise improve the answers in knowledge graphs, store provenance information in knowledge graphs, and augment the knowledge graphs using other data sources. Thus, the reliability and correctness of chatbots will be improved as the bugs and inaccuracies in answers provided by the LLM will be corrected in the knowledge graphs, but the chatbots can still harness the abilities of LLMs to provide answers across various subject-matter domains.

公开(公告)号：US11743122B1

公开(公告)日：2023-08-29

申请号：US17709068

申请日：2022-03-30

Applicant: Amazon Technologies, Inc.

Inventor： Samuel Bayless ,  John David Backes ,  Daniel William Dacosta ,  Vaibhav Katkade ,  Sagar Chintamani Joshi ,  Nadia Labai ,  Syed Mubashir Iqbal ,  Patrick Trentin ,  Nathan Launchbury ,  Nikolaos Giannarakis ,  Victor Heorhiadi ,  Nick Matthews

IPC: H04L41/0869 ,  H04L41/08 ,  H04L41/22 ,  H04L41/0816 ,  H04L41/14 ,  H04L41/147 ,  H04L9/40

CPC classification number: H04L41/0869 ,  H04L41/0816 ,  H04L41/0883 ,  H04L41/145 ,  H04L41/147 ,  H04L41/22 ,  H04L63/0263

Abstract: A network change verification (NCV) system is disclosed for checking whether a proposed configuration change on a network alters the way that the network controls recently observed network flows. In embodiments, the system builds an observed flow control model (OFCM) from logs of recent flows observed in the network. The OFCM, which may be periodically updated based on newly observed flows, provides a compact representation of how individual network flows were ostensibly controlled by the network. When a proposed configuration change is received, the system analyzes the change against the OFCM to check whether the change will alter how the network controls recently observed flows. If so, the proposed change is blocked, and an alert is generated identifying flows that are affected by the change. The NCV system thus prevents network operators from accidentally making changes on the network that will materially alter the behavior of the network.