{"id":48076,"date":"2023-12-08T09:04:39","date_gmt":"2023-12-08T08:04:39","guid":{"rendered":"https:\/\/www.inovex.de\/?p=48076"},"modified":"2023-12-08T09:04:39","modified_gmt":"2023-12-08T08:04:39","slug":"how-to-leverage-knowledge-graphs-in-question-answering","status":"publish","type":"post","link":"https:\/\/www.inovex.de\/de\/blog\/how-to-leverage-knowledge-graphs-in-question-answering\/","title":{"rendered":"How to Leverage Knowledge Graphs in Question Answering?"},"content":{"rendered":"<p><span style=\"font-weight: 400;\">Knowledge Graphs (KGs) have emerged as valuable repositories of world knowledge. The integration of KGs with Natural Language Processing (NLP) models has opened new avenues for applications such as question answering<\/span><span style=\"font-weight: 400;\">, text classification<\/span><span style=\"font-weight: 400;\">, text generation<\/span><span style=\"font-weight: 400;\">, and machine translation<\/span><span style=\"font-weight: 400;\">, leading to remarkable performance gains.<\/span> <span style=\"font-weight: 400;\">In this blog post, we explore the methodologies for integrating KGs into Question-Answering (QA) models.<\/span><!--more--><\/p>\n<p>We begin with an introduction to knowledge graphs. Next, we summarize existing methods for integrating KGs into NLP models. Following that, we explain the task of question answering and\u00a0introduce our novel KG-enhanced model for generative QA that utilizes Graph Neural Networks (GNNs). Finally, we conclude the article with some key insights from our study.<\/p>\n<div id=\"ez-toc-container\" class=\"ez-toc-v2_0_83 counter-hierarchy ez-toc-counter ez-toc-custom ez-toc-container-direction\">\n<div class=\"ez-toc-title-container\"><p class=\"ez-toc-title\" style=\"cursor:inherit\"><\/p>\n<\/div><nav><ul class='ez-toc-list ez-toc-list-level-1 ' ><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-1\" href=\"https:\/\/www.inovex.de\/de\/blog\/how-to-leverage-knowledge-graphs-in-question-answering\/#Knowledge-Graphs\" >Knowledge Graphs<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-2\" href=\"https:\/\/www.inovex.de\/de\/blog\/how-to-leverage-knowledge-graphs-in-question-answering\/#Methods-for-Integrating-KG-into-NLP-Models\" >Methods for Integrating KG into NLP Models<\/a><ul class='ez-toc-list-level-3' ><li class='ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-3\" href=\"https:\/\/www.inovex.de\/de\/blog\/how-to-leverage-knowledge-graphs-in-question-answering\/#1-KG-to-Text\" >1. KG-to-Text<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-4\" href=\"https:\/\/www.inovex.de\/de\/blog\/how-to-leverage-knowledge-graphs-in-question-answering\/#2-Designing-KG-aware-pre-training-tasks\" >2. Designing KG-aware pre-training tasks<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-3'><a class=\"ez-toc-link ez-toc-heading-5\" href=\"https:\/\/www.inovex.de\/de\/blog\/how-to-leverage-knowledge-graphs-in-question-answering\/#3-Build-a-knowledge-module-into-the-model\" >3. Build a knowledge module into the model<\/a><\/li><\/ul><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-6\" href=\"https:\/\/www.inovex.de\/de\/blog\/how-to-leverage-knowledge-graphs-in-question-answering\/#Question-Answering\" >Question Answering<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-7\" href=\"https:\/\/www.inovex.de\/de\/blog\/how-to-leverage-knowledge-graphs-in-question-answering\/#The-proposed-KG-enhances-QA-model\" >The proposed KG-enhances QA model<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-8\" href=\"https:\/\/www.inovex.de\/de\/blog\/how-to-leverage-knowledge-graphs-in-question-answering\/#Evaluation\" >Evaluation<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-9\" href=\"https:\/\/www.inovex.de\/de\/blog\/how-to-leverage-knowledge-graphs-in-question-answering\/#Conclusion\" >Conclusion<\/a><\/li><\/ul><\/nav><\/div>\n<h2><span class=\"ez-toc-section\" id=\"Knowledge-Graphs\"><\/span>Knowledge Graphs<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>Knowledge graphs are powerful representations of information in graph-like structures, serving as rich, interconnected knowledge representations. A knowledge graph consists of nodes and edges. Nodes symbolize entities, which could be real-world objects, individuals, or abstract concepts. Edges denote relationships between entities.<\/p>\n<div>\n<dl id=\"attachment_48097\">\n<dt>\n<p><figure id=\"attachment_48097\" aria-describedby=\"caption-attachment-48097\" style=\"width: 775px\" class=\"wp-caption aligncenter\"><img loading=\"lazy\" decoding=\"async\" class=\"wp-image-48097\" src=\"https:\/\/www.inovex.de\/wp-content\/uploads\/ontology-1.jpg\" alt=\"Knowledge Graph with nodes that are connected by edges\" width=\"775\" height=\"615\" srcset=\"https:\/\/www.inovex.de\/wp-content\/uploads\/ontology-1.jpg 899w, https:\/\/www.inovex.de\/wp-content\/uploads\/ontology-1-300x238.jpg 300w, https:\/\/www.inovex.de\/wp-content\/uploads\/ontology-1-768x610.jpg 768w, https:\/\/www.inovex.de\/wp-content\/uploads\/ontology-1-400x318.jpg 400w, https:\/\/www.inovex.de\/wp-content\/uploads\/ontology-1-360x286.jpg 360w\" sizes=\"auto, (max-width: 775px) 100vw, 775px\" \/><figcaption id=\"caption-attachment-48097\" class=\"wp-caption-text\">An example of a knowledge graph. Credit: https:\/\/timbr.ai\/blog\/introducing-timbr-sql-kg\/<\/figcaption><\/figure><\/dt>\n<\/dl>\n<\/div>\n<p>For example, in a social network graph, nodes signify people and edges represent relationships like spouse, parent, or friend. A node can also represent a broader category of other nodes, for instance, all individuals have an \u201cis a\u201c\u00a0relationship with the \u201cperson\u201c\u00a0node, allowing KGs to capture hierarchical relationships between entities. The flexibility of KGs allows them to represent a wide range of information from basic statements like \u201cBerlin is the capital city of Germany\u201c\u00a0to more intricate, qualified assertions such as \u201call capitals are cities.\u201c\u00a0A knowledge graph is often represented as a set of triples (h, r, t) where h, r, and t represent the head entity, relation, and tail entity, respectively. For example, (Berlin, capital city of, Germany), (Berlin, is a, city), (Steve Jobs, former CEO of, Apple), and (Steve Jobs, is a, person).<\/p>\n<p>Representing information as a knowledge graph offers several advantages. One key advantage is the graph&#8217;s inherent ability to capture complex relationships between concepts, providing a rich and structured representation of knowledge. Another benefit is flexibility and extensibility. KGs allow systematic integration of new data sources into the existing graph. This enables KGs to grow and evolve alongside expanding knowledge bases and changing requirements. Moreover, knowledge graphs can also efficiently integrate diverse types of data, including text, images, audio, videos, and ontologies, into a unified and cohesive representation. Additionally, graph-based representations facilitate efficient querying, reasoning, advanced analytics, and the derivation of deeper insights from the data.<\/p>\n<p>One of the frequently used knowledge graphs in NLP research is <a href=\"https:\/\/static.googleusercontent.com\/media\/research.google.com\/en\/\/pubs\/archive\/42240.pdf\" target=\"_blank\" rel=\"noopener\">Wikidata<\/a>, a large-scale general domain KG. Wikidata covers a wide range of knowledge, from famous people, events, music, and movies, to chemical substances. Wikidata has proven beneficial for open-domain QA. Another prominent KG in NLP research is Conceptnet, which focuses on common sense knowledge. Conceptnet is found to be useful for common sense reasoning tasks.<\/p>\n<h2><span class=\"ez-toc-section\" id=\"Methods-for-Integrating-KG-into-NLP-Models\"><\/span>Methods for Integrating KG into NLP Models<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>In recent years, there has been a growing effort to leverage the structured and semantically rich nature of knowledge graphs to enhance the capabilities of NLP models, especially pre-training language models (LM) and QA systems. KGs are regarded as valuable sources of factual knowledge which is crucial for QA tasks. Effectively integrating KG knowledge gives rise to two challenges. Firstly, how to identify relevant knowledge to extract from the KG. Secondly, how to find the optimal approach to inject this knowledge into NLP systems to maximize its potential while maintaining efficiency. Common types of knowledge extracted from KGs are entities, triples, and subgraphs. In the pre-train and fine-tune framework, KG knowledge can be integrated during the pre-training phase of a language model, the fine-tuning phase, or both phases. Methods for knowledge fusion are, for instance, converting KG knowledge into text (KG-to-text), designing KG-aware pre-training tasks for the language model, and incorporating a knowledge module into the NLP model.<\/p>\n<h3><span class=\"ez-toc-section\" id=\"1-KG-to-Text\"><\/span>1. KG-to-Text<span class=\"ez-toc-section-end\"><\/span><\/h3>\n<div>\n<dl id=\"attachment_48138\">\n<dt><img loading=\"lazy\" decoding=\"async\" class=\"size-full wp-image-48100 aligncenter\" src=\"https:\/\/www.inovex.de\/wp-content\/uploads\/kg_to_text.png\" alt=\"An illustration of the kg-to-text approach\" width=\"1177\" height=\"707\" srcset=\"https:\/\/www.inovex.de\/wp-content\/uploads\/kg_to_text.png 1177w, https:\/\/www.inovex.de\/wp-content\/uploads\/kg_to_text-300x180.png 300w, https:\/\/www.inovex.de\/wp-content\/uploads\/kg_to_text-1024x615.png 1024w, https:\/\/www.inovex.de\/wp-content\/uploads\/kg_to_text-768x461.png 768w, https:\/\/www.inovex.de\/wp-content\/uploads\/kg_to_text-400x240.png 400w, https:\/\/www.inovex.de\/wp-content\/uploads\/kg_to_text-360x216.png 360w\" sizes=\"auto, (max-width: 1177px) 100vw, 1177px\" \/><\/dt>\n<\/dl>\n<\/div>\n<p>The most straightforward strategy to leverage KG knowledge is to convert KG knowledge (e.g., subgraphs or triples) into natural language sentences and use them as supplementary contextual information. This conversion can be achieved through various techniques, including template-based methods, pre-training or fine-tuning language models on KG-to-text datasets, and using GNNs as graph encoders.<\/p>\n<p>Note: The illustration is taken from <a href=\"https:\/\/aclanthology.org\/2021.naacl-main.278\/\" target=\"_blank\" rel=\"noopener\">Agarwal et al. 2021<\/a> and has been modified.<\/p>\n<h3><span class=\"ez-toc-section\" id=\"2-Designing-KG-aware-pre-training-tasks\"><\/span>2. Designing KG-aware pre-training tasks<span class=\"ez-toc-section-end\"><\/span><\/h3>\n<p>Given the widespread use of pre-trained language models in NLP, one knowledge fusion strategy is to pre-train the language models not only on text corpora but also on knowledge graphs. Typical approaches involve using entity-aware and relation-aware pre-training objectives in conjunction with the language model\u2019s original objectives, such as Masked Language Modeling (MLM). The goal is to map knowledge from both text corpora and knowledge graphs into the same embedding space during pre-training.<\/p>\n<p>A popular entity-aware objective involves predicting masked entity spans. Another objective is the entity linking task. Given a specific token span, the model is asked to predict the corresponding KG entity.<\/p>\n<div>\n<dl id=\"attachment_48102\">\n<dt>\n<p><figure id=\"attachment_48102\" aria-describedby=\"caption-attachment-48102\" style=\"width: 842px\" class=\"wp-caption aligncenter\"><img loading=\"lazy\" decoding=\"async\" class=\"wp-image-48102 size-full\" src=\"https:\/\/www.inovex.de\/wp-content\/uploads\/entity_predition_task.png\" alt=\"Masked Entity Spans: Entity spans arrow to the Language Model that arrows to words\" width=\"842\" height=\"227\" srcset=\"https:\/\/www.inovex.de\/wp-content\/uploads\/entity_predition_task.png 842w, https:\/\/www.inovex.de\/wp-content\/uploads\/entity_predition_task-300x81.png 300w, https:\/\/www.inovex.de\/wp-content\/uploads\/entity_predition_task-768x207.png 768w, https:\/\/www.inovex.de\/wp-content\/uploads\/entity_predition_task-400x108.png 400w, https:\/\/www.inovex.de\/wp-content\/uploads\/entity_predition_task-360x97.png 360w\" sizes=\"auto, (max-width: 842px) 100vw, 842px\" \/><figcaption id=\"caption-attachment-48102\" class=\"wp-caption-text\">The illustration depicts an entity span prediction task, where the model aims to predict the masked entity span.<\/figcaption><\/figure><\/dt>\n<\/dl>\n<\/div>\n<p>Another effective approach is the replacement-detection objective. Here, entity spans in the text are replaced with random entities, and the model learns to identify whether the entity spans have been replaced.<\/p>\n<div>\n<dl id=\"attachment_48104\">\n<dt>\n<p><figure id=\"attachment_48104\" aria-describedby=\"caption-attachment-48104\" style=\"width: 837px\" class=\"wp-caption aligncenter\"><img loading=\"lazy\" decoding=\"async\" class=\"wp-image-48104 size-full\" src=\"https:\/\/www.inovex.de\/wp-content\/uploads\/entity_replacement_prediction_task.png\" alt=\"Overview of replacement-detection objective: entity spans arrow to language model that arrows to checks or crosses\" width=\"837\" height=\"222\" srcset=\"https:\/\/www.inovex.de\/wp-content\/uploads\/entity_replacement_prediction_task.png 837w, https:\/\/www.inovex.de\/wp-content\/uploads\/entity_replacement_prediction_task-300x80.png 300w, https:\/\/www.inovex.de\/wp-content\/uploads\/entity_replacement_prediction_task-768x204.png 768w, https:\/\/www.inovex.de\/wp-content\/uploads\/entity_replacement_prediction_task-400x106.png 400w, https:\/\/www.inovex.de\/wp-content\/uploads\/entity_replacement_prediction_task-360x95.png 360w\" sizes=\"auto, (max-width: 837px) 100vw, 837px\" \/><figcaption id=\"caption-attachment-48104\" class=\"wp-caption-text\">The illustration depicts the entity replacement prediction task, where the model predicts whether a token span (corresponding to an entity) has been replaced by another token span (an incorrect entity, here &#8222;Steven Spielberg&#8220;).<\/figcaption><\/figure><\/dt>\n<\/dl>\n<\/div>\n<p>Relation-aware objectives involve, for example, predicting the tail entity given the head entity and the relation. While KG-aware pre-training has the drawback of being computationally expensive and time-consuming, this approach enables the language model to gain a deep understanding of both textual and knowledge-based information.<\/p>\n<p>Note: The illustrations are taken from <a href=\"https:\/\/arxiv.org\/pdf\/2012.03551.pdf\" target=\"_blank\" rel=\"noopener\">He et al. 2020<\/a> and have been modified.<\/p>\n<h3><span class=\"ez-toc-section\" id=\"3-Build-a-knowledge-module-into-the-model\"><\/span>3. Build a knowledge module into the model<span class=\"ez-toc-section-end\"><\/span><\/h3>\n<p>A knowledge module aims to provide KG knowledge to the primary NLP model. This module can be modeled as an ad-hoc component separately from the primary model or can be integrated into the model. For instance, by inserting it between the model\u2019s layers, or by utilizing it in the calculation of attention scores.<\/p>\n<div>\n<dl id=\"attachment_48110\">\n<dt>\n<p><figure id=\"attachment_48110\" aria-describedby=\"caption-attachment-48110\" style=\"width: 481px\" class=\"wp-caption aligncenter\"><img loading=\"lazy\" decoding=\"async\" class=\" wp-image-48110\" src=\"https:\/\/www.inovex.de\/wp-content\/uploads\/add_knowledge_modul_on_top.png\" alt=\"Buttons of Language Model and Knowledge Graph point to knowledge integration module points to prediction\" width=\"481\" height=\"284\" srcset=\"https:\/\/www.inovex.de\/wp-content\/uploads\/add_knowledge_modul_on_top.png 827w, https:\/\/www.inovex.de\/wp-content\/uploads\/add_knowledge_modul_on_top-300x177.png 300w, https:\/\/www.inovex.de\/wp-content\/uploads\/add_knowledge_modul_on_top-768x453.png 768w, https:\/\/www.inovex.de\/wp-content\/uploads\/add_knowledge_modul_on_top-400x236.png 400w, https:\/\/www.inovex.de\/wp-content\/uploads\/add_knowledge_modul_on_top-360x212.png 360w\" sizes=\"auto, (max-width: 481px) 100vw, 481px\" \/><figcaption id=\"caption-attachment-48110\" class=\"wp-caption-text\">Building a knowledge module on top of the pre-trained LM<\/figcaption><\/figure><\/dt>\n<dd>\n<p><figure id=\"attachment_48112\" aria-describedby=\"caption-attachment-48112\" style=\"width: 501px\" class=\"wp-caption aligncenter\"><img loading=\"lazy\" decoding=\"async\" class=\" wp-image-48112\" src=\"https:\/\/www.inovex.de\/wp-content\/uploads\/insert_knowledge_module.png\" alt=\"Graph: Language Model Layer and seperate knowledge graph point to Knowledge integration Layer that points to language model layer\" width=\"501\" height=\"195\" srcset=\"https:\/\/www.inovex.de\/wp-content\/uploads\/insert_knowledge_module.png 958w, https:\/\/www.inovex.de\/wp-content\/uploads\/insert_knowledge_module-300x117.png 300w, https:\/\/www.inovex.de\/wp-content\/uploads\/insert_knowledge_module-768x299.png 768w, https:\/\/www.inovex.de\/wp-content\/uploads\/insert_knowledge_module-400x156.png 400w, https:\/\/www.inovex.de\/wp-content\/uploads\/insert_knowledge_module-360x140.png 360w\" sizes=\"auto, (max-width: 501px) 100vw, 501px\" \/><figcaption id=\"caption-attachment-48112\" class=\"wp-caption-text\"><strong>Inserting a knowledge module between LM\u2019s layers<\/strong><\/figcaption><\/figure><\/dd>\n<\/dl>\n<\/div>\n<h2><span class=\"ez-toc-section\" id=\"Question-Answering\"><\/span>Question Answering<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>Question answering is an NLP task in which the system is given a question in natural language and asked to provide an answer to the question. QA systems have attracted a lot of attention in both research and industry due to their ability to facilitate information search. There is a wide range of applications in real-world scenarios, from supporting virtual assistants such as Siri and Google Assistant to improving customer support in the form of QA chatbots.<\/p>\n<p>QA systems can be categorized into open-book and close-book settings. In the open-book setting, QA systems are allowed to access external information to support the response process, for example, by retrieving data from databases or websites. In a close-book system, the QA system cannot access external sources of information. QA systems that operate in a close-book system tend to be very large and require a tedious training procedure because they need to contain all knowledge within their parameters. A good example of such a system is ChatGPT. Nowadays, the \u201cretriever-reader\u201c architecture is considered one of the most efficient and promising approaches. The retriever&#8217;s task is to find relevant documents for a given question. The reader is responsible for deriving the final answer from the retrieved documents.<\/p>\n<h2><span class=\"ez-toc-section\" id=\"The-proposed-KG-enhances-QA-model\"><\/span>The proposed KG-enhances QA model<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>Our proposed model adopts the third method of knowledge fusion, which involves using a separate knowledge module with the QA model where knowledge fusion happens in the input layer of the QA model. The selected form of knowledge is subgraph, and the fusion step takes place during the fine-tuning stage. The model consists of two main components: the QA module and the GNN module.<\/p>\n<figure id=\"attachment_48168\" aria-describedby=\"caption-attachment-48168\" style=\"width: 1485px\" class=\"wp-caption aligncenter\"><img loading=\"lazy\" decoding=\"async\" class=\"size-full wp-image-48168\" src=\"https:\/\/www.inovex.de\/wp-content\/uploads\/model-5.png\" alt=\"The proposed KG-enhances QA model\" width=\"1485\" height=\"861\" srcset=\"https:\/\/www.inovex.de\/wp-content\/uploads\/model-5.png 1485w, https:\/\/www.inovex.de\/wp-content\/uploads\/model-5-300x174.png 300w, https:\/\/www.inovex.de\/wp-content\/uploads\/model-5-1024x594.png 1024w, https:\/\/www.inovex.de\/wp-content\/uploads\/model-5-768x445.png 768w, https:\/\/www.inovex.de\/wp-content\/uploads\/model-5-400x232.png 400w, https:\/\/www.inovex.de\/wp-content\/uploads\/model-5-360x209.png 360w\" sizes=\"auto, (max-width: 1485px) 100vw, 1485px\" \/><figcaption id=\"caption-attachment-48168\" class=\"wp-caption-text\"><strong>The architecture of our knowledge graph enhanced QA model. z denotes the question node and p denotes the pool graph vector. A token sequence consists of a question and a passage.<\/strong><\/figcaption><\/figure>\n<p>The QA module is built on a generative QA model called Fusion in Decoder (FiD) proposed by <a href=\"https:\/\/aclanthology.org\/2021.eacl-main.74\/\" target=\"_blank\" rel=\"noopener\">Izacard and Grave 2020<\/a> at Facebook AI Research. FiD is an encoder-decoder model based on the pre-trained T5 model. Its strength lies in its ability to handle a large number of supporting passages (up to 100 passages). Our GNN module is derived from a powerful GNN-based reasoning module of another QA model, called QA-GNN. QA-GNN uses KG subgraphs as a knowledge source and encodes them using a Graph Attention Network (<a href=\"https:\/\/arxiv.org\/abs\/1710.10903\" target=\"_blank\" rel=\"noopener\">GAT<\/a>) to improve the reasoning ability of the model. GAT is a variation of GNN that incorporates the attention mechanism. The authors modified GAT so that it can incorporate information from the relationships and node scores. They also proposed a method to assign a relevance score to each node, which helps in selecting important nodes for the subgraphs. However, QA-GNN is only applicable for multi-choice QA scenarios. Recognizing the potential benefit of the GNN module in QA-GNN, we wished to expand its applicability to generative QA.<\/p>\n<p>Our model can be regarded as a reader system since we provide the model with support passages. The original FiD model encodes one passage at a time, then concatenates all encoded passages into a long vector sequence and passes it to the decoder to generate the answer. To incorporate KG knowledge, we simply combine KG embeddings with token embeddings into the same sequence at the input layer. Specifically, for each question, we first create a subgraph based on KG entities found in the question and all passages. Similar to QA-GNN, we add a special node into the subgraph which represents the question, and connect it to other nodes through a newly added relation. We also expand the subgraph by adding some relevant 1-hop neighboring nodes that form a path within the subgraph. This results in a fully connected subgraph. We initialize nodes in the subgraphs with pre-trained embeddings. After that, we apply five rounds of GNN. The message-passing process of GNN is supposed to allow the entities to interact with the question and learn about the relationship between them. Then, we take the learned node embeddings, the pooled graph embeddings, and the question node embeddings, and concatenate them with the token embeddings of the question and a passage. The token embeddings are obtained through the embedding layer of the pre-trained T5 encoder. While FiD only uses token embeddings as input, we use token embeddings and also KG embeddings.<\/p>\n<p>Our idea is to allow token and KG entities to exchange information in the encoder through multiple layers of attention. After encoding, we obtain a sequence of embeddings (vectors) for each passage. To reduce the computing time and memory space in the decoder, we remove the embeddings associated with the subgraph and keep only the token embeddings.\u00a0 After doing the same for all passages, we concatenate the remaining token embeddings from each passage together, resulting in a long vector sequence. Finally, we pass it as input to the decoder to generate the answer.<\/p>\n<p>One point to note about this architecture is that we put token and KG embeddings in the same sequence. This usually requires an alignment to transform KG embeddings into the same space as token embeddings. In our work, the message-passing in GNN serves as an alignment procedure since the subgraph consists of both KG embeddings and LM embeddings through the added question node. After applying GNN, we further align the embeddings using a linear transformation.<\/p>\n<h2><span class=\"ez-toc-section\" id=\"Evaluation\"><\/span>Evaluation<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>We evaluate the model on a small subset of <a href=\"https:\/\/nlp.cs.washington.edu\/triviaqa\/\" target=\"_blank\" rel=\"noopener\">TriviaQA<\/a> and the GQA (Graph Question-Answering) dataset. GQA is a private QA dataset that provides high-quality Wikidata subgraphs for each question. Both datasets cover general topics (open domain). We experiment with two knowledge graphs, <a href=\"https:\/\/github.com\/THU-KEG\/KEPLER\" target=\"_blank\" rel=\"noopener\">Wikidata5m <\/a>and <a href=\"https:\/\/conceptnet.io\/\" target=\"_blank\" rel=\"noopener\">Conceptnet<\/a>.<\/p>\n<p>An example of the TriviaQA dataset<\/p>\n<div class=\"bg-color-blue\">\n<p><b>Question:<\/b><span style=\"font-weight: 400;\"> The Dodecanese Campaign of WWII that was an attempt by the Allied forces to capture islands in the Aegean Sea was the inspiration for which acclaimed 1961 commando film?<\/span><\/p>\n<p><b>Answer:<\/b><span style=\"font-weight: 400;\"> The Guns of Navarone<\/span><\/p>\n<p><b>Passage: <\/b><span style=\"font-weight: 400;\">The Dodecanese Campaign of World War II was an attempt by Allied forces to capture the Italian-held Dodecanese islands in the Aegean Sea following the surrender of Italy in September 1943, and use them as bases against the German-controlled Balkans. The failed campaign, and in particular the Battle of Leros, inspired the 1957 novel <\/span><b>The Guns of Navarone<\/b><span style=\"font-weight: 400;\"> and the successful 1961 movie of the same name.<\/span><\/p>\n<\/div>\n<p><b>An example of the GQA dataset<\/b><\/p>\n<div class=\"bg-color-blue\">\n<p><b>Question:<\/b><span style=\"font-weight: 400;\"> What American Mexican cinematographer did Brokeback Mountain star<\/span><span style=\"font-weight: 400;\"><br \/>\n<\/span><b>Answer:<\/b><span style=\"font-weight: 400;\"> Rodrigo Prieto<\/span><\/p>\n<p><b>Subgraph:<\/b><\/p>\n<p><img loading=\"lazy\" decoding=\"async\" class=\"aligncenter wp-image-48115 \" src=\"https:\/\/www.inovex.de\/wp-content\/uploads\/subgraph-1024x544.png\" alt=\"\" width=\"524\" height=\"278\" srcset=\"https:\/\/www.inovex.de\/wp-content\/uploads\/subgraph-1024x544.png 1024w, https:\/\/www.inovex.de\/wp-content\/uploads\/subgraph-300x160.png 300w, https:\/\/www.inovex.de\/wp-content\/uploads\/subgraph-768x408.png 768w, https:\/\/www.inovex.de\/wp-content\/uploads\/subgraph-400x213.png 400w, https:\/\/www.inovex.de\/wp-content\/uploads\/subgraph-360x191.png 360w, https:\/\/www.inovex.de\/wp-content\/uploads\/subgraph.png 1057w\" sizes=\"auto, (max-width: 524px) 100vw, 524px\" \/><\/p>\n<\/div>\n<p>In all experiments, our model outperforms the baseline which is the original FiD model. Overall, the model shows an improvement gain of 0.06 to 1.44% in <a href=\"https:\/\/aclanthology.org\/2021.eacl-main.74\/\" target=\"_blank\" rel=\"noopener\">Exact Match<\/a> scores (EM). The performance gain on TriviaQA is not significant, while it is significant on GQA. We find that our model is most beneficial in a few-shot setting when we train the model with 10 samples (+1.44% EM).<\/p>\n<p>Although the model can provide a certain performance gain, it comes with longer processing times that are nearly double when compared to the baseline FiD. This is possibly due to the input sequence becoming longer because of the added KG embeddings, and the computation time for self-attention in the encoder growing quadratically with the sequence length. Another factor is the need to run the GNN on the fly for each question. The computation time scales with the subgraph&#8217;s size, the dimension of the node embedding, and the number of GNN layers.\u00a0In addition, the construction of subgraphs is the most time-consuming and memory-intensive step. This step involves identifying KG entities in the input text and computing a relevant score for the entities. Entity linking requires careful configuration because words in the text can be ambiguous and a KG entity can have many aliases. For a large KG like Wikidata5m, the entity linking process consumes a significant amount of time. This was the reason why we could not use the whole TriviaQA dataset, but decided to work with a subset instead. Similarly, the node scoring demands substantial processing time as we extend the subgraph to neighboring nodes. It requires us to calculate a score for all child nodes of the initial subgraph before we can rank and choose the relevant nodes to build the final subgraphs. In the case of Wikidata5m, some parent nodes can have more than 3,000 child nodes.<\/p>\n<h2><span class=\"ez-toc-section\" id=\"Conclusion\"><\/span>Conclusion<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>The rising popularity of knowledge graphs in NLP research can be attributed to their ability to represent information in a structured way, preserving the connectivity of real-world entities and concepts. Moreover, knowledge graphs can effectively integrate diverse data modalities and sources into a unified representation and can be systematically extended and updated. As the era of big data evolves, with information constantly changing, the need for reliable, fact-aware NLP systems becomes crucial. In this context, knowledge graphs have the potential to serve as dynamic sources of factual knowledge, enhancing accuracy and contextual understanding in NLP tasks.<\/p>\n<p>In this blog post, we presented different techniques for integrating knowledge graphs into NLP models with a focus on question-answering tasks. To gain a practical understanding, we developed a KG-enhanced question-answering model based on a simple approach. The model exhibited performance improvements in both fully supervised and few-shot scenarios across two KGs and two QA datasets. However, it has limitations in terms of efficiency.<\/p>\n<p>While knowledge graphs are attractive sources of valuable information, integrating them into NLP models is challenging due to the need to optimize various interconnected components. For instance, within our model architecture, we had to look for effective methods for identifying KG entities in the input text, accurately estimating node relevance, finding the optimal strategy for constructing subgraphs and designing a suitable knowledge fusion approach.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Knowledge Graphs (KGs) have emerged as valuable repositories of world knowledge. The integration of KGs with Natural Language Processing (NLP) models has opened new avenues for applications such as question answering, text classification, text generation, and machine translation, leading to remarkable performance gains. In this blog post, we explore the methodologies for integrating KGs into [&hellip;]<\/p>\n","protected":false},"author":293,"featured_media":49994,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"inline_featured_image":false,"ep_exclude_from_search":false,"footnotes":""},"tags":[511,151,577,569,379],"service":[76,75],"coauthors":[{"id":293,"display_name":"Suteera Seeha","user_nicename":"sseeha"}],"class_list":["post-48076","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","tag-artificial-intelligence-2","tag-deep-learning","tag-graph-neural-networks","tag-knowledge-graph","tag-nlu","service-artificial-intelligence","service-nlp"],"acf":[],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v27.7 - https:\/\/yoast.com\/product\/yoast-seo-wordpress\/ -->\n<title>How to Leverage Knowledge Graphs in Question Answering? - inovex GmbH<\/title>\n<meta name=\"description\" content=\"In this blog post, we explore methods for integrating knowledge graphs into Question Answering (QA) models and\u00a0introduce our novel KG-enhanced model for generative QA that utilizes Graph Neural Networks.\" \/>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/www.inovex.de\/de\/blog\/how-to-leverage-knowledge-graphs-in-question-answering\/\" \/>\n<meta property=\"og:locale\" content=\"de_DE\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"How to Leverage Knowledge Graphs in Question Answering? - inovex GmbH\" \/>\n<meta property=\"og:description\" content=\"In this blog post, we explore methods for integrating knowledge graphs into Question Answering (QA) models and\u00a0introduce our novel KG-enhanced model for generative QA that utilizes Graph Neural Networks.\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.inovex.de\/de\/blog\/how-to-leverage-knowledge-graphs-in-question-answering\/\" \/>\n<meta property=\"og:site_name\" content=\"inovex GmbH\" \/>\n<meta property=\"article:publisher\" content=\"https:\/\/www.facebook.com\/inovexde\" \/>\n<meta property=\"article:published_time\" content=\"2023-12-08T08:04:39+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/www.inovex.de\/wp-content\/uploads\/Blogheader-Knowledge-Graphs.png\" \/>\n\t<meta property=\"og:image:width\" content=\"1500\" \/>\n\t<meta property=\"og:image:height\" content=\"880\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/png\" \/>\n<meta name=\"author\" content=\"Suteera Seeha\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:image\" content=\"https:\/\/www.inovex.de\/wp-content\/uploads\/Blogheader-Knowledge-Graphs-1024x601.png\" \/>\n<meta name=\"twitter:creator\" content=\"@inovexgmbh\" \/>\n<meta name=\"twitter:site\" content=\"@inovexgmbh\" \/>\n<meta name=\"twitter:label1\" content=\"Verfasst von\" \/>\n\t<meta name=\"twitter:data1\" content=\"Suteera Seeha\" \/>\n\t<meta name=\"twitter:label2\" content=\"Gesch\u00e4tzte Lesezeit\" \/>\n\t<meta name=\"twitter:data2\" content=\"13\u00a0Minuten\" \/>\n\t<meta name=\"twitter:label3\" content=\"Written by\" \/>\n\t<meta name=\"twitter:data3\" content=\"Suteera Seeha\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"Article\",\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/blog\\\/how-to-leverage-knowledge-graphs-in-question-answering\\\/#article\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/blog\\\/how-to-leverage-knowledge-graphs-in-question-answering\\\/\"},\"author\":{\"name\":\"Suteera Seeha\",\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/#\\\/schema\\\/person\\\/361a5df2031157225ea4a94a6cbdfeae\"},\"headline\":\"How to Leverage Knowledge Graphs in Question Answering?\",\"datePublished\":\"2023-12-08T08:04:39+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/blog\\\/how-to-leverage-knowledge-graphs-in-question-answering\\\/\"},\"wordCount\":2672,\"commentCount\":0,\"publisher\":{\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/#organization\"},\"image\":{\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/blog\\\/how-to-leverage-knowledge-graphs-in-question-answering\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/www.inovex.de\\\/wp-content\\\/uploads\\\/Blogheader-Knowledge-Graphs.png\",\"keywords\":[\"Artificial Intelligence\",\"Deep Learning\",\"Graph Neural Networks\",\"Knowledge Graph\",\"NLU\"],\"articleSection\":[\"Analytics\",\"English Content\",\"General\"],\"inLanguage\":\"de\",\"potentialAction\":[{\"@type\":\"CommentAction\",\"name\":\"Comment\",\"target\":[\"https:\\\/\\\/www.inovex.de\\\/de\\\/blog\\\/how-to-leverage-knowledge-graphs-in-question-answering\\\/#respond\"]}]},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/blog\\\/how-to-leverage-knowledge-graphs-in-question-answering\\\/\",\"url\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/blog\\\/how-to-leverage-knowledge-graphs-in-question-answering\\\/\",\"name\":\"How to Leverage Knowledge Graphs in Question Answering? - inovex GmbH\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/blog\\\/how-to-leverage-knowledge-graphs-in-question-answering\\\/#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/blog\\\/how-to-leverage-knowledge-graphs-in-question-answering\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/www.inovex.de\\\/wp-content\\\/uploads\\\/Blogheader-Knowledge-Graphs.png\",\"datePublished\":\"2023-12-08T08:04:39+00:00\",\"description\":\"In this blog post, we explore methods for integrating knowledge graphs into Question Answering (QA) models and\u00a0introduce our novel KG-enhanced model for generative QA that utilizes Graph Neural Networks.\",\"breadcrumb\":{\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/blog\\\/how-to-leverage-knowledge-graphs-in-question-answering\\\/#breadcrumb\"},\"inLanguage\":\"de\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/www.inovex.de\\\/de\\\/blog\\\/how-to-leverage-knowledge-graphs-in-question-answering\\\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"de\",\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/blog\\\/how-to-leverage-knowledge-graphs-in-question-answering\\\/#primaryimage\",\"url\":\"https:\\\/\\\/www.inovex.de\\\/wp-content\\\/uploads\\\/Blogheader-Knowledge-Graphs.png\",\"contentUrl\":\"https:\\\/\\\/www.inovex.de\\\/wp-content\\\/uploads\\\/Blogheader-Knowledge-Graphs.png\",\"width\":1500,\"height\":880,\"caption\":\"Grafische Darstellung eines Netzwerks mit mehreren Personen, die miteinander verbunden sind. In der Mitte befindet sich ein Laptop, umgeben von Datenbl\u00f6cken. Weitere Ger\u00e4te wie ein Smartphone sind ebenfalls abgebildet.\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/blog\\\/how-to-leverage-knowledge-graphs-in-question-answering\\\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"How to Leverage Knowledge Graphs in Question Answering?\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/#website\",\"url\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/\",\"name\":\"inovex GmbH\",\"description\":\"\",\"publisher\":{\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"de\"},{\"@type\":\"Organization\",\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/#organization\",\"name\":\"inovex GmbH\",\"url\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"de\",\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/#\\\/schema\\\/logo\\\/image\\\/\",\"url\":\"https:\\\/\\\/www.inovex.de\\\/wp-content\\\/uploads\\\/2021\\\/03\\\/inovex-logo-16-9-1.png\",\"contentUrl\":\"https:\\\/\\\/www.inovex.de\\\/wp-content\\\/uploads\\\/2021\\\/03\\\/inovex-logo-16-9-1.png\",\"width\":1921,\"height\":1081,\"caption\":\"inovex GmbH\"},\"image\":{\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/#\\\/schema\\\/logo\\\/image\\\/\"},\"sameAs\":[\"https:\\\/\\\/www.facebook.com\\\/inovexde\",\"https:\\\/\\\/x.com\\\/inovexgmbh\",\"https:\\\/\\\/www.instagram.com\\\/inovexlife\\\/\",\"https:\\\/\\\/www.linkedin.com\\\/company\\\/inovex\",\"https:\\\/\\\/www.youtube.com\\\/channel\\\/UC7r66GT14hROB_RQsQBAQUQ\"]},{\"@type\":\"Person\",\"@id\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/#\\\/schema\\\/person\\\/361a5df2031157225ea4a94a6cbdfeae\",\"name\":\"Suteera Seeha\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"de\",\"@id\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/f100ab39473fc0755243b406bfab68ad324f453b5583adc016ec89ffdddb9a6a?s=96&d=retro&r=g095283dcf5e4e92f88549871bb1fef68\",\"url\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/f100ab39473fc0755243b406bfab68ad324f453b5583adc016ec89ffdddb9a6a?s=96&d=retro&r=g\",\"contentUrl\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/f100ab39473fc0755243b406bfab68ad324f453b5583adc016ec89ffdddb9a6a?s=96&d=retro&r=g\",\"caption\":\"Suteera Seeha\"},\"url\":\"https:\\\/\\\/www.inovex.de\\\/de\\\/blog\\\/author\\\/sseeha\\\/\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"How to Leverage Knowledge Graphs in Question Answering? - inovex GmbH","description":"In this blog post, we explore methods for integrating knowledge graphs into Question Answering (QA) models and\u00a0introduce our novel KG-enhanced model for generative QA that utilizes Graph Neural Networks.","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/www.inovex.de\/de\/blog\/how-to-leverage-knowledge-graphs-in-question-answering\/","og_locale":"de_DE","og_type":"article","og_title":"How to Leverage Knowledge Graphs in Question Answering? - inovex GmbH","og_description":"In this blog post, we explore methods for integrating knowledge graphs into Question Answering (QA) models and\u00a0introduce our novel KG-enhanced model for generative QA that utilizes Graph Neural Networks.","og_url":"https:\/\/www.inovex.de\/de\/blog\/how-to-leverage-knowledge-graphs-in-question-answering\/","og_site_name":"inovex GmbH","article_publisher":"https:\/\/www.facebook.com\/inovexde","article_published_time":"2023-12-08T08:04:39+00:00","og_image":[{"width":1500,"height":880,"url":"https:\/\/www.inovex.de\/wp-content\/uploads\/Blogheader-Knowledge-Graphs.png","type":"image\/png"}],"author":"Suteera Seeha","twitter_card":"summary_large_image","twitter_image":"https:\/\/www.inovex.de\/wp-content\/uploads\/Blogheader-Knowledge-Graphs-1024x601.png","twitter_creator":"@inovexgmbh","twitter_site":"@inovexgmbh","twitter_misc":{"Verfasst von":"Suteera Seeha","Gesch\u00e4tzte Lesezeit":"13\u00a0Minuten","Written by":"Suteera Seeha"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"Article","@id":"https:\/\/www.inovex.de\/de\/blog\/how-to-leverage-knowledge-graphs-in-question-answering\/#article","isPartOf":{"@id":"https:\/\/www.inovex.de\/de\/blog\/how-to-leverage-knowledge-graphs-in-question-answering\/"},"author":{"name":"Suteera Seeha","@id":"https:\/\/www.inovex.de\/de\/#\/schema\/person\/361a5df2031157225ea4a94a6cbdfeae"},"headline":"How to Leverage Knowledge Graphs in Question Answering?","datePublished":"2023-12-08T08:04:39+00:00","mainEntityOfPage":{"@id":"https:\/\/www.inovex.de\/de\/blog\/how-to-leverage-knowledge-graphs-in-question-answering\/"},"wordCount":2672,"commentCount":0,"publisher":{"@id":"https:\/\/www.inovex.de\/de\/#organization"},"image":{"@id":"https:\/\/www.inovex.de\/de\/blog\/how-to-leverage-knowledge-graphs-in-question-answering\/#primaryimage"},"thumbnailUrl":"https:\/\/www.inovex.de\/wp-content\/uploads\/Blogheader-Knowledge-Graphs.png","keywords":["Artificial Intelligence","Deep Learning","Graph Neural Networks","Knowledge Graph","NLU"],"articleSection":["Analytics","English Content","General"],"inLanguage":"de","potentialAction":[{"@type":"CommentAction","name":"Comment","target":["https:\/\/www.inovex.de\/de\/blog\/how-to-leverage-knowledge-graphs-in-question-answering\/#respond"]}]},{"@type":"WebPage","@id":"https:\/\/www.inovex.de\/de\/blog\/how-to-leverage-knowledge-graphs-in-question-answering\/","url":"https:\/\/www.inovex.de\/de\/blog\/how-to-leverage-knowledge-graphs-in-question-answering\/","name":"How to Leverage Knowledge Graphs in Question Answering? - inovex GmbH","isPartOf":{"@id":"https:\/\/www.inovex.de\/de\/#website"},"primaryImageOfPage":{"@id":"https:\/\/www.inovex.de\/de\/blog\/how-to-leverage-knowledge-graphs-in-question-answering\/#primaryimage"},"image":{"@id":"https:\/\/www.inovex.de\/de\/blog\/how-to-leverage-knowledge-graphs-in-question-answering\/#primaryimage"},"thumbnailUrl":"https:\/\/www.inovex.de\/wp-content\/uploads\/Blogheader-Knowledge-Graphs.png","datePublished":"2023-12-08T08:04:39+00:00","description":"In this blog post, we explore methods for integrating knowledge graphs into Question Answering (QA) models and\u00a0introduce our novel KG-enhanced model for generative QA that utilizes Graph Neural Networks.","breadcrumb":{"@id":"https:\/\/www.inovex.de\/de\/blog\/how-to-leverage-knowledge-graphs-in-question-answering\/#breadcrumb"},"inLanguage":"de","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.inovex.de\/de\/blog\/how-to-leverage-knowledge-graphs-in-question-answering\/"]}]},{"@type":"ImageObject","inLanguage":"de","@id":"https:\/\/www.inovex.de\/de\/blog\/how-to-leverage-knowledge-graphs-in-question-answering\/#primaryimage","url":"https:\/\/www.inovex.de\/wp-content\/uploads\/Blogheader-Knowledge-Graphs.png","contentUrl":"https:\/\/www.inovex.de\/wp-content\/uploads\/Blogheader-Knowledge-Graphs.png","width":1500,"height":880,"caption":"Grafische Darstellung eines Netzwerks mit mehreren Personen, die miteinander verbunden sind. In der Mitte befindet sich ein Laptop, umgeben von Datenbl\u00f6cken. Weitere Ger\u00e4te wie ein Smartphone sind ebenfalls abgebildet."},{"@type":"BreadcrumbList","@id":"https:\/\/www.inovex.de\/de\/blog\/how-to-leverage-knowledge-graphs-in-question-answering\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/www.inovex.de\/de\/"},{"@type":"ListItem","position":2,"name":"How to Leverage Knowledge Graphs in Question Answering?"}]},{"@type":"WebSite","@id":"https:\/\/www.inovex.de\/de\/#website","url":"https:\/\/www.inovex.de\/de\/","name":"inovex GmbH","description":"","publisher":{"@id":"https:\/\/www.inovex.de\/de\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/www.inovex.de\/de\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"de"},{"@type":"Organization","@id":"https:\/\/www.inovex.de\/de\/#organization","name":"inovex GmbH","url":"https:\/\/www.inovex.de\/de\/","logo":{"@type":"ImageObject","inLanguage":"de","@id":"https:\/\/www.inovex.de\/de\/#\/schema\/logo\/image\/","url":"https:\/\/www.inovex.de\/wp-content\/uploads\/2021\/03\/inovex-logo-16-9-1.png","contentUrl":"https:\/\/www.inovex.de\/wp-content\/uploads\/2021\/03\/inovex-logo-16-9-1.png","width":1921,"height":1081,"caption":"inovex GmbH"},"image":{"@id":"https:\/\/www.inovex.de\/de\/#\/schema\/logo\/image\/"},"sameAs":["https:\/\/www.facebook.com\/inovexde","https:\/\/x.com\/inovexgmbh","https:\/\/www.instagram.com\/inovexlife\/","https:\/\/www.linkedin.com\/company\/inovex","https:\/\/www.youtube.com\/channel\/UC7r66GT14hROB_RQsQBAQUQ"]},{"@type":"Person","@id":"https:\/\/www.inovex.de\/de\/#\/schema\/person\/361a5df2031157225ea4a94a6cbdfeae","name":"Suteera Seeha","image":{"@type":"ImageObject","inLanguage":"de","@id":"https:\/\/secure.gravatar.com\/avatar\/f100ab39473fc0755243b406bfab68ad324f453b5583adc016ec89ffdddb9a6a?s=96&d=retro&r=g095283dcf5e4e92f88549871bb1fef68","url":"https:\/\/secure.gravatar.com\/avatar\/f100ab39473fc0755243b406bfab68ad324f453b5583adc016ec89ffdddb9a6a?s=96&d=retro&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/f100ab39473fc0755243b406bfab68ad324f453b5583adc016ec89ffdddb9a6a?s=96&d=retro&r=g","caption":"Suteera Seeha"},"url":"https:\/\/www.inovex.de\/de\/blog\/author\/sseeha\/"}]}},"_links":{"self":[{"href":"https:\/\/www.inovex.de\/de\/wp-json\/wp\/v2\/posts\/48076","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.inovex.de\/de\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.inovex.de\/de\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.inovex.de\/de\/wp-json\/wp\/v2\/users\/293"}],"replies":[{"embeddable":true,"href":"https:\/\/www.inovex.de\/de\/wp-json\/wp\/v2\/comments?post=48076"}],"version-history":[{"count":6,"href":"https:\/\/www.inovex.de\/de\/wp-json\/wp\/v2\/posts\/48076\/revisions"}],"predecessor-version":[{"id":49903,"href":"https:\/\/www.inovex.de\/de\/wp-json\/wp\/v2\/posts\/48076\/revisions\/49903"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.inovex.de\/de\/wp-json\/wp\/v2\/media\/49994"}],"wp:attachment":[{"href":"https:\/\/www.inovex.de\/de\/wp-json\/wp\/v2\/media?parent=48076"}],"wp:term":[{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.inovex.de\/de\/wp-json\/wp\/v2\/tags?post=48076"},{"taxonomy":"service","embeddable":true,"href":"https:\/\/www.inovex.de\/de\/wp-json\/wp\/v2\/service?post=48076"},{"taxonomy":"author","embeddable":true,"href":"https:\/\/www.inovex.de\/de\/wp-json\/wp\/v2\/coauthors?post=48076"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}