KTRL+F: Knowledge-Augmented In-Document Search
Abstract: We introduce a new problem KTRL+F, a knowledge-augmented in-document search task that necessitates real-time identification of all semantic targets within a document with the awareness of external sources through a single natural query. KTRL+F addresses following unique challenges for in-document search: 1)utilizing knowledge outside the document for extended use of additional information about targets, and 2) balancing between real-time applicability with the performance. We analyze various baselines in KTRL+F and find limitations of existing models, such as hallucinations, high latency, or difficulties in leveraging external knowledge. Therefore, we propose a Knowledge-Augmented Phrase Retrieval model that shows a promising balance between speed and performance by simply augmenting external knowledge in phrase embedding. We also conduct a user study to verify whether solving KTRL+F can enhance search experience for users. It demonstrates that even with our simple model, users can reduce the time for searching with less queries and reduced extra visits to other sources for collecting evidence. We encourage the research community to work on KTRL+F to enhance more efficient in-document information access.
- Unlimiformer: Long-range transformers with unlimited length input. arXiv.
- Annotating documents with relevant wikipedia concepts. In SiKDD.
- Language models are few-shot learners. In NeurIPS.
- Medusa: Simple llm inference acceleration framework with multiple decoding heads. arXiv.
- Vicuna: An open-source chatbot impressing gpt-4 with 90%* chatgpt quality.
- Jacob Cohen. 1960. A coefficient of agreement for nominal scales. Educational and Psychological Measurement.
- Quoref: A reading comprehension dataset with questions requiring coreferential reasoning. In EMNLP.
- A dataset of information-seeking questions and answers anchored in research papers. In NAACL.
- Autoregressive entity retrieval. arXiv.
- Bert: Pre-training of deep bidirectional transformers for language understanding. In NAACL.
- Iirc: A dataset of incomplete information reading comprehension questions. In EMNLP.
- Entities as experts: Sparse memory access with entity supervision. In EMNLP.
- MRQA 2019 shared task: Evaluating generalization in reading comprehension. In Proceedings of the 2nd Workshop on Machine Reading for Question Answering.
- Joseph L. Fleiss. 1971. Measuring nominal scale agreement among many raters. Psychological Bulletin.
- Break the sequential dependency of llm inference using lookahead decoding. arXiv.
- Debertav3: Improving deberta using electra-style pre-training with gradient-disentangled embedding sharing. In ICLR.
- Gautier Izacard and Edouard Grave. 2021. Leveraging passage retrieval with generative models for open domain question answering. In EACL.
- TriviaQA: A large scale distantly supervised challenge dataset for reading comprehension. In ACL.
- Challenges and applications of large language models. arXiv.
- Omar Khattab and Matei Zaharia. 2020. Colbert: Efficient and effective passage search via contextualized late interaction over bert. In SIGIR.
- Natural questions: A benchmark for question answering research. TACL.
- Efficient memory management for large language model serving with pagedattention. In SIGOPSSIGOPS.
- J Richard Landis and Gary G. Koch. 1977. The measurement of observer agreement for categorical data. Biometrics.
- Nonparametric decoding for generative retrieval. In Findings of ACL.
- Learning dense representations of phrases at scale. In ACL.
- Sensebert: Driving some sense into bert. In ACL.
- Multispanqa: A dataset for multi-span question answering. In ACL.
- Aggretriever: A simple approach to aggregate textual representations for robust dense passage retrieval. TACL.
- OpenAI. 2022. Chatgpt: Optimizing language models for dialogue.
- OpenAI. 2023. Gpt-4 technical report.
- Check your facts and try again: Improving large language models with external knowledge and automated feedback. arXiv.
- Knowledge enhanced contextual word representations. In EMNLP-IJCNLP.
- E-bert: Efficient-yet-effective entity embeddings for bert. In Findings of EMNLP.
- Exploring the limits of transfer learning with a unified text-to-text transformer. JMLR.
- Erate: Efficient retrieval augmented text embeddings. In First Workshop on Insights from Negative Results in NLP.
- Know what you don’t know: Unanswerable questions for SQuAD. In ACL.
- Squad: 100,000+ questions for machine comprehension of text. In EMNLP.
- In-context retrieval-augmented language models. TACL.
- How much knowledge can you pack into the parameters of a language model? In EMNLP.
- ColBERTv2: Effective and efficient retrieval via lightweight late interaction. In NAACL.
- A simple and effective model for answering multi-span questions. In EMNLP.
- Llama 2: Open foundation and fine-tuned chat models. arXiv.
- NewsQA: A machine comprehension dataset. In Proceedings of the 2nd Workshop on Representation Learning for NLP.
- K-adapter: Infusing knowledge into pre-trained models with adapters. In Findings of ACL-IJCNLP.
- Scalable zero-shot entity linking with dense entity retrieval. In EMNLP.
- Pretrained encyclopedia: Weakly supervised knowledge-pretrained language model. arXiv.
- Generate rather than retrieve: Large language models are strong context generators. In ICLR.
- Ernie: Enhanced language representation with informative entities. In ACL.
- Romqa: A benchmark for robust, multi-evidence, multi-answer question answering. In Findings of EMNLP.
- Question answering with long multiple-span answers. In EMNLP.
Paper Prompts
Sign up for free to create and run prompts on this paper using GPT-5.
Top Community Prompts
Collections
Sign up for free to add this paper to one or more collections.