More Like this

1. Wafer Map Pattern Analytics Driven By Natural Language Queries

   https://doi.org/10.1109/ITCAsia55616.2022.00016  

   Zeng, Yueling Jenny; Yang, Min Jian; Wang, Li-C. (August 2022, 2022 IEEE International Test Conference in Asia (ITC-Asia))

   We present a novel approach where wafer map pattern analytics are driven by natural language queries. At the core is a semantic parser that translates a user query into a meaning representation comprising instructions to generate a summary plot. The allowable plot types are pre-defined which serve as an interface that communicates user intents to the analytics software backend. Application results on wafer maps from a recent production line are presented to explain the capabilities and benefits of the proposed approach. 

   more » « less
2. Learning A Wafer Feature With One Training Sample

   https://doi.org/10.1109/ITC44778.2020.9325254  

   Zeng, Yueling Jenny; Wang, Li-C.; Shan, Chuanhe Jay; Sumikawa, Nik (November 2020, 2020 IEEE International Test Conference (ITC))

   null (Ed.)

   In this work, we consider learning a wafer plot recognizer where only one training sample is available. We introduce an approach called Manifestation Learning to enable the learning. The underlying technology utilizes the Variational AutoEncoder (VAE) approach to construct a so-called Manifestation Space. The training sample is projected into this space and the recognition is achieved through a pre-trained model in the space. Using wafer probe test data from an automotive product line, this paper explains the learning approach, its feasibility and limitation. 

   more » « less
3. Energon: A Data Acquisition System for Portable Building Analytics

   https://doi.org/10.1145/3447555.3464850  

   He, Fang; Deng, Yang; Xu, Yanhui; Xu, Cheng; Hong, Dezhi; Wang, Dan (June 2021, e-Energy)

   Emerging building analytics rely on data-driven machine learning algorithms. However, writing these analytics is still challenging— developers need to know not only what data are required by the analytics but also how to reach the data in each individual building, despite the existing solutions to standardizing data and resource management in buildings. To bridge the gap between analytics development and the specific details of reaching actual data in each building, we present Energon, an open-source system that enables portable building analytics. The core of Energon is a new data organization for building as well as tools that can effectively manage building data and support building analytics development. More specifically, we propose a new "logic partition" of data resources in buildings, and this abstraction universally applies to all buildings. We develop a declarative query language accordingly to f ind data resources in this new logic view with high-level queries, thus substantially reducing development efforts. We also develop a query engine with automatic data extraction by traversing building ontology that widely exists in buildings. In this way, Energon enables mapping of analytics requirements to building resources in a building-agnostic manner. Using four types of real-world building analytics, we demonstrate the use of Energon and its effectiveness in reducing development efforts. 

   more » « less
4. Large Language Model Augmented Narrative Driven Recommendations

   https://doi.org/10.1145/3604915  

   Mysore, Sheshera; McCallum, Andrew; Zamani, Hamed (September 2023, ACM)

   Zhang, Jie; Chen, Li; Berkovsky, Shlomo; Zhang, Min; Noia, Tommaso di; Basilico, Justin; Pizzato, Luiz; Song, Yang (Ed.)

   Narrative-driven recommendation (NDR) presents an information access problem where users solicit recommendations with verbose descriptions of their preferences and context, for example, travelers soliciting recommendations for points of interest while describing their likes/dislikes and travel circumstances. These requests are increasingly important with the rise of natural language-based conversational interfaces for search and recommendation systems. However, NDR lacks abundant training data for models, and current platforms commonly do not support these requests. Fortunately, classical user-item interaction datasets contain rich textual data, e.g., reviews, which often describe user preferences and context – this may be used to bootstrap training for NDR models. In this work, we explore using large language models (LLMs) for data augmentation to train NDR models. We use LLMs for authoring synthetic narrative queries from user-item interactions with few-shot prompting and train retrieval models for NDR on synthetic queries and user-item interaction data. Our experiments demonstrate that this is an effective strategy for training small-parameter retrieval models that outperform other retrieval and LLM baselines for narrative-driven recommendation. 

   more » « less
5. Large Language Model Augmented Narrative Driven Recommendations

   https://doi.org/10.1145/3604915.3608829  

   Mysore, Sheshera; Mccallum, Andrew; Zamani, Hamed (September 2023, ACM)

   Narrative-driven recommendation (NDR) presents an information access problem where users solicit recommendations with verbose descriptions of their preferences and context, for example, travelers soliciting recommendations for points of interest while describing their likes/dislikes and travel circumstances. These requests are increasingly important with the rise of natural language-based conversational interfaces for search and recommendation systems. However, NDR lacks abundant training data for models, and current platforms commonly do not support these requests. Fortunately, classical user-item interaction datasets contain rich textual data, e.g., reviews, which often describe user preferences and context – this may be used to bootstrap training for NDR models. In this work, we explore using large language models (LLMs) for data augmentation to train NDR models. We use LLMs for authoring synthetic narrative queries from user-item interactions with few-shot prompting and train retrieval models for NDR on synthetic queries and user-item interaction data. Our experiments demonstrate that this is an effective strategy for training small-parameter retrieval models that outperform other retrieval and LLM baselines for narrative-driven recommendation. 

   more » « less