Rachith Aiyappa

Year	Rank	Type	Title / Venue / Authors
2026	J	`jnl`	Zero-shot stance detection in practice: insights on training, prompting, and decoding with a capable lightweight LLM. PeerJ Comput. Sci. Rachith Aiyappa, Shruthi Senthilmani, Jisun An, Haewoon Kwak, Yong-Yeol Ahn
2025	J	`jnl`	A Benchmark for Zero-Shot Belief Inference in Large Language Models. CoRR Joseph Malone, Rachith Aiyappa, Byunghwee Lee, Haewoon Kwak, Jisun An, Yong-Yeol Ahn
2025	A*	`conf`	Implicit degree bias in the link prediction task. ICML Rachith Aiyappa, Xin Wang, Munjung Kim, Ozgur Can Seckin, Yong-Yeol Ahn, Sadamori Kojaku
2024	J	`jnl`	Benchmarking zero-shot stance detection with FlanT5-XXL: Insights from training data, prompting, and decoding strategies into its near-SoTA performance. CoRR Rachith Aiyappa, Shruthi Senthilmani, Jisun An, Haewoon Kwak, Yong-Yeol Ahn
2024	J	`jnl`	Implicit degree bias in the link prediction task. CoRR Rachith Aiyappa, Xin Wang, Munjung Kim, Ozgur Can Seckin, Jisung Yoon, Yong-Yeol Ahn, Sadamori Kojaku
2024	J	`jnl`	Information diffusion assumptions can distort our understanding of social network dynamics. CoRR Matthew R. DeVerna, Francesco Pierri, Rachith Aiyappa, Diogo Pacheco, John Bryden, Filippo Menczer
2024	J	`jnl`	Neural embedding of beliefs reveals the role of relative dissonance in human decision-making. CoRR Byunghwee Lee, Rachith Aiyappa, Yong-Yeol Ahn, Haewoon Kwak, Jisun An
2023	A	`conf`	A Multi-Platform Collection of Social Media Posts about the 2022 U.S. Midterm Elections. ICWSM Rachith Aiyappa, Matthew R. DeVerna, Manita Pote, Bao Tran Truong, Wanying Zhao, David Axelrod, Aria Pessianzadeh, Zoher Kachwala, Munjung Kim, Ozgur Can Seckin, Minsuk Kim, Sunny Gandhi, Amrutha Manikonda, Francesco Pierri, Filippo Menczer, Kai-Cheng Yang
2023	J	`jnl`	A multi-platform collection of social media posts about the 2022 U.S. midterm elections. CoRR Matthew R. DeVerna, Rachith Aiyappa, Wanying Zhao, Manita Pote, Bao Tran Truong, David Axelrod, Aria Pessianzadeh, Zoher Kachwala, Munjung Kim, Ozgur Can Seckin, Minsuk Kim, Sunny Gandhi, Amrutha Manikonda, Francesco Pierri, Filippo Menczer, Kai-Cheng Yang
2023	J	`jnl`	Can we trust the evaluation on ChatGPT? CoRR Rachith Aiyappa, Jisun An, Haewoon Kwak, Yong-Yeol Ahn
2023	J	`jnl`	The Inexplicable Efficacy of Language Models. XRDS Rachith Aiyappa, Zoher Kachwala
2023	J	`jnl`	Weighted Belief Networks Unify Simple and Complex Contagion Dynamics. CoRR Rachith Aiyappa, Alessandro Flammini, Yong-Yeol Ahn
2022	J	`jnl`	Identification and characterization of misinformation superspreaders on social media. CoRR Matthew R. DeVerna, Rachith Aiyappa, Diogo Pacheco, John Bryden, Filippo Menczer

from hashlib import md5, sha1, sha256
import inspect
from datetime import datetime, timezone
from typing import Any

import magic
from magika import Magika

from redb.extractors.enum import Tag
from redb.extractors.pe_extractor import PEExtractor
from redb.models.dataclasses import PEOverlay


class PEOverlayExtractor(PEExtractor):

    def __init__(
        self,
        filepath,
        log,
        exporters=None,
        index_prefix=None,
        elastic_index=None,
        known_benign=False,
        known_malicious=False,
        pe=None,
    ):
        super().__init__(
            filepath,
            log,
            exporters,
            index_prefix,
            elastic_index,
            known_benign,
            known_malicious,
            pe,
        )
        self.elastic_index = self.index_prefix + "-pe_overlay"
        self.log.debug(inspect.currentframe().f_code.co_name)

    def tag(self):
        return Tag.PE_OVERLAY.value

    def _extract_pe_overlay(self):
        self.log.debug(inspect.currentframe().f_code.co_name)
        overlay = self.pe.get_overlay()
        if overlay:
            return PEOverlay(
                _id=sha256(overlay).hexdigest(),
                overlay_size=len(overlay),
                overlay_entropy=self.calculate_entropy(overlay),
                overlay_offset=self.pe.get_overlay_data_start_offset(),
                overlay_mimetype=magic.from_buffer(overlay, mime=True),
                overlay_type=magic.from_buffer(overlay),
                overlay_magika=Magika().identify_bytes(overlay).output.ct_label,
                overlay_sha256=sha256(overlay).hexdigest(),
                overlay_sha1=sha1(overlay).hexdigest(),
                overlay_md5=md5(overlay).hexdigest(),
            )
        return None

    def extract(self):
        try:
            self.log.debug(inspect.currentframe().f_code.co_name)
            overlay = self._extract_pe_overlay()
            # self.export_to_elastic([overlay])  # Let the exporters handle this
            return overlay
        except Exception as e:
            self.log.error(f"Error extracting PE overlay: {e}")
            return None

    def prepare_export_data(self, exporter_type: str) -> Any:
        if exporter_type == "ElasticsearchExporter":
            return self.extract()
        elif exporter_type == "ClickHouseExporter":
            overlay = self.extract()
            if overlay is None:
                return None
            
            data = []
            current_time = datetime.now(timezone.utc)
            
            data.append([
                self.sha256,              # sha256
                self.md5,                 # md5
                self.sha1,                # sha1
                overlay.overlay_size,     # overlay_size
                overlay.overlay_entropy,  # overlay_entropy
                overlay.overlay_offset,   # overlay_offset
                overlay.overlay_mimetype, # overlay_mimetype
                overlay.overlay_type,     # overlay_type
                overlay.overlay_magika,   # overlay_magika
                overlay.overlay_sha256,   # overlay_sha256
                overlay.overlay_sha1,     # overlay_sha1
                overlay.overlay_md5,      # overlay_md5
                current_time             # analysis_date
            ])
            
            column_names = [
                'sha256', 'md5', 'sha1', 'overlay_size', 'overlay_entropy',
                'overlay_offset', 'overlay_mimetype', 'overlay_type', 'overlay_magika',
                'overlay_sha256', 'overlay_sha1', 'overlay_md5',
                'analysis_date'
            ]
            
            if not data:
                return None

            column_type_names = [
                'FixedString(64)', 'FixedString(32)', 'FixedString(40)',
                'UInt64', 'Float64', 'UInt64',
                'LowCardinality(String)', 'LowCardinality(String)', 'LowCardinality(String)',
                'FixedString(64)', 'FixedString(40)', 'FixedString(32)',
                'DateTime64(3, \'UTC\')'
            ]

            return (data, column_names, column_type_names)

    def get_clickhouse_table(self) -> str:
        return "redb_pe_overlays"