Threshold Influence Model for Allocating Advertising Budgets
Authors: Atsushi Miyauchi, Yuni Iwamasa, Takuro Fukunaga, Naonori Kakimura
ICML 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We conduct thorough experiments to confirm that our algorithms outperform baseline algorithms. |
| Researcher Affiliation | Academia | Atsushi Miyauchi MIYAUCHI.A.AA@M.TITECH.AC.JP Graduate School of Decision Science and Technology, Tokyo Institute of Technology, Japan Yuni Iwamasa YUNI IWAMASA@MIST.I.U-TOKYO.AC.JP Graduate School of Information Science and Technology, University of Tokyo, Japan Takuro Fukunaga TAKURO@NII.AC.JP National Institute of Informatics, JST, ERATO, Kawarabayashi Large Graph Project, Japan Naonori Kakimura KAKIMURA@GLOBAL.C.U-TOKYO.AC.JP Graduate School of Arts and Sciences, University of Tokyo, Japan |
| Pseudocode | Yes | Algorithm 1 INCREMENTALGREEDY |
| Open Source Code | No | The paper does not provide explicit statements or links indicating that the source code for their methodology is publicly available. |
| Open Datasets | Yes | As a real-world network, we use the Yahoo! Search Marketing Advertiser Bidding Data (Yahoo!) as in (Soma et al., 2014). [...] Yahoo! Webscope dataset: A1 Yahoo! Search Marketing Advertiser Bidding Data, version 1.0. http://webscope.sandbox.yahoo.com/ catalog.php?datatype=a. |
| Dataset Splits | No | The paper describes the datasets used and how certain parameters are set (e.g., thresholds), but it does not specify any training, validation, or test dataset splits, percentages, or explicit sample counts. |
| Hardware Specification | Yes | All experiments were performed on a Windows PC with Intel Core i7 2.40 GHz CPU and 16 GB RAM. |
| Software Dependencies | No | The paper mentions that 'Our algorithms were implemented in C++.' but does not list specific versions for compilers, libraries, or other software dependencies. |
| Experiment Setup | Yes | A sequence of probabilities in the functions is set randomly from the interval [0, 0.1]. We define cs = 1 for each source s and wt = 1 for each target t. We set us = 1 as we can reduce any instance into the case. We set threshold θt for each target t in the following two ways: random thresholds means θt is chosen randomly from the interval [0, 1], and large thresholds means θt is chosen randomly from the interval [0.5, 1]. We set |S| = 20,000 and |T| = 200,000 with around 2,000,000 edges between them. The graphs are generated so that the degree distribution of the source nodes obeys the power law with exponent γ = 2.0. After assigning the degree deg(s) to each source node s, it is connected to deg(s) nodes chosen uniformly from T. We set θt = 0.8 for every target t. |