Tutorial on Heterogeneous Entity Matching

Outline
1. Entity Matching
      - Definition and goal
      - Process of entity matching
      - Classification of methods
2. Understanding Heterogeneous Entity Matching
      - Defining the concept of heterogeneity
      - Cross-lingual matching, cross-domain and multimodal entity linking
3. Challenges of Heterogeneous Entity Matching
4. Advanced Methods for Heterogeneous Entity Matching
      - Active learning, learning by example, and crowdsourcing
      - Transfer learning and domain adaptation
      - Incorporating external knowledge sources
5. Applications of Heterogeneous Entity Matching
6. Challenges and Opportunities for Future Research

Presenters:
 -Behshid Behkamal, Assistant Professor, Department of Computer Engineering, Ferdowsi University of Mashhad, Iran
 -Mostafa Milani, Assistant Professor, Department of Computer Science at the Univesity of Western Ontario, Canada

Duration:  4 Hours

                                                               Prompt engineering

Abstract:

In recent years, the growing artificial intelligence community has made great efforts to develop large-scale artificial intelligence models (LAMs) through the massive increase in data and computing resources.The technique of prompting has emerged as a crucial tool in harnessing the diverse capabilities of these AI models. By utilizing prompts, we can effectively steer the behavior of large AI models through explicit commands and illustrative examples. In this workshop, we will discuss the basic concepts of Large Language Models(LLMs) and prompting and various techniques for creating effective prompts. we introduce the OpenAI library and showcase examples of prompting techniques implemented using this library. We define prompt engineering and explore different methods of prompt engineering, specifically focusing on soft prompts and hard prompts. We introduce OpenPrompt, an open-source tool for prompt engineering. Then we showcase real-world applications where prompts have proven valuable in solving NLP problems

Presenters:

-Razie Hashemi, Master's Student in Computer Engineering, Department of Computer Engineering, Ferdowsi University of Mashhad, Iran

-Fatemeh Rahim Farkhani, Master's Student in Computer Engineering, Department of Computer Engineering, Ferdowsi University of Mashhad, Iran

Duration: 4 Hours

                                                         Computational psychology

Part 1:

Abstract:
In recent years, computational psychology has gained special popularity as one of the emerging fields among computer and psychology researchers. Extracting hidden knowledge from the mass of data recorded by psychologists by modern tools invented in computer science and artificial intelligence has increased the speed and accurate diagnosis and prediction of mental disorders.
In this regard, the "Artificial Intelligence in Health" research, Ferdowsi University's web technology laboratory, in line with the latest developments, in recent years, with the guidance of respected professors and the efforts of graduate students, has achieved valuable success in this field. It is hoped that by proposing this issue, an opportunity will be provided for more acquaintance and researchers interested in this field.
Headlines:
     - Introduction to computational psychology
     - Applications of computational psychology
     - Analysis of social networks to diagnose and predict mental disorders
     - Introducing the eRisk challenge

Part2:

Abstract:

Adverse drug reactions (ADE) are complications that occur when taking drugs in normal doses. ADE is a public health problem, as it hospitalizes millions of patients worldwide each year. Early detection of ADE reduces economic costs and prevents deaths. Diagnosis of adverse drug reactions usually depends on voluntary reporting or medical information. But in recent years, the data sent by the user in social networks has become an important source for this work. Limiting the number of words on Twitter allows users to use words in a targeted and focused way. User-provided information about drugs and their adverse effects on Twitter is an important resource for post-marketing drug monitoring. Vaccine has been one of the most successful public health interventions to date. However, vaccines are medicinal products that have risks, so that many side effects are reported after receiving the vaccine. Traditional adverse event reporting systems suffer from several critical challenges, including timeliness and inefficiency. This increases the motivation of social media-based diagnosis systems, which demonstrate the successful ability to obtain timely and prevalent disease information. Social media can receive epidemic information from social sensors because they reflect public mood and trends that can be used to identify vaccine side effects. In the last decade, researches conducted in the field of identifying side effects of drugs and vaccines have widely used this valuable data source. In this research, we examine the research done in the field of identifying the side effects of drugs and vaccines.

Titles:

     - An introduction to the extraction of drug and vaccine side effects from social networks
     - Reviewing the researches done in this field
     - Challenges in this field

Presenters:

- Reza saedi, PhD student in computer engineering, Department of Computer Engineering, Ferdowsi University of Mashhad, Iran

- fariba mohammadi-khah, Master of Computer Engineering, Department of Computer Engineering, Ferdowsi University of Mashhad, Iran