Details
| Originalsprache | Englisch |
|---|---|
| Titel des Sammelwerks | WWW 2015 Companion - Proceedings of the 24th International Conference on World Wide Web |
| Seiten | 1339-1344 |
| Seitenumfang | 6 |
| ISBN (elektronisch) | 9781450334730 |
| Publikationsstatus | Veröffentlicht - 18 Mai 2015 |
| Veranstaltung | 24th International Conference on World Wide Web, WWW 2015 - Florence, Italien Dauer: 18 Mai 2015 → 22 Mai 2015 |
Abstract
In many cases, a user turns to search engines to find information about real-world situations, namely, political elections, sport competitions, or natural disasters. Such temporal querying behavior can be observed through a significant number of event-related queries generated in web search. In this paper, we study the task of detecting event-related queries, which is the first step for understanding temporal query intent and enabling different temporal search applications, e.g., time-aware query auto-completion, temporal ranking, and result diversi cation. We propose a two-step approach to detecting events from query logs. We first identify a set of event candidates by considering both implicit and explicit temporal information needs. The next step further classi es the candidates into two main categories, namely, event or non-event. In more detail, we leverage different machine learning techniques for query classi- cation, which are trained using the feature set composed of time series features from signal processing, along with features derived from click-through information, and standard statistical features. In order to evaluate our proposed approach, we conduct an experiment using two real-world query logs with manually annotated relevance assessments for 837 events. To this end, we provide a large set of eventrelated queries made available for fostering research on this challenging task.
ASJC Scopus Sachgebiete
- Informatik (insg.)
- Computernetzwerke und -kommunikation
- Informatik (insg.)
- Software
Zitieren
- Standard
- Harvard
- Apa
- Vancouver
- BibTex
- RIS
WWW 2015 Companion - Proceedings of the 24th International Conference on World Wide Web. 2015. S. 1339-1344.
Publikation: Beitrag in Buch/Bericht/Sammelwerk/Konferenzband › Aufsatz in Konferenzband › Forschung › Peer-Review
}
TY - GEN
T1 - Learning to Detect Event-Related Queries for Web Search
AU - Kanhabua, Nattiya
AU - Nguyen, Tu Ngoc
AU - Nejdl, Wolfgang
N1 - Funding information: This work was partially funded by the European Commission for the FP7 project ForgetIT and the ERC Advanced Grant ALEXANDRIA under the grant numbers 600826 and 339233, respectively.
PY - 2015/5/18
Y1 - 2015/5/18
N2 - In many cases, a user turns to search engines to find information about real-world situations, namely, political elections, sport competitions, or natural disasters. Such temporal querying behavior can be observed through a significant number of event-related queries generated in web search. In this paper, we study the task of detecting event-related queries, which is the first step for understanding temporal query intent and enabling different temporal search applications, e.g., time-aware query auto-completion, temporal ranking, and result diversi cation. We propose a two-step approach to detecting events from query logs. We first identify a set of event candidates by considering both implicit and explicit temporal information needs. The next step further classi es the candidates into two main categories, namely, event or non-event. In more detail, we leverage different machine learning techniques for query classi- cation, which are trained using the feature set composed of time series features from signal processing, along with features derived from click-through information, and standard statistical features. In order to evaluate our proposed approach, we conduct an experiment using two real-world query logs with manually annotated relevance assessments for 837 events. To this end, we provide a large set of eventrelated queries made available for fostering research on this challenging task.
AB - In many cases, a user turns to search engines to find information about real-world situations, namely, political elections, sport competitions, or natural disasters. Such temporal querying behavior can be observed through a significant number of event-related queries generated in web search. In this paper, we study the task of detecting event-related queries, which is the first step for understanding temporal query intent and enabling different temporal search applications, e.g., time-aware query auto-completion, temporal ranking, and result diversi cation. We propose a two-step approach to detecting events from query logs. We first identify a set of event candidates by considering both implicit and explicit temporal information needs. The next step further classi es the candidates into two main categories, namely, event or non-event. In more detail, we leverage different machine learning techniques for query classi- cation, which are trained using the feature set composed of time series features from signal processing, along with features derived from click-through information, and standard statistical features. In order to evaluate our proposed approach, we conduct an experiment using two real-world query logs with manually annotated relevance assessments for 837 events. To this end, we provide a large set of eventrelated queries made available for fostering research on this challenging task.
KW - Events
KW - Query Intent
KW - Temporal Queryspecification
UR - http://www.scopus.com/inward/record.url?scp=84936983785&partnerID=8YFLogxK
U2 - 10.1145/2740908.2741698
DO - 10.1145/2740908.2741698
M3 - Conference contribution
AN - SCOPUS:84936983785
SP - 1339
EP - 1344
BT - WWW 2015 Companion - Proceedings of the 24th International Conference on World Wide Web
T2 - 24th International Conference on World Wide Web, WWW 2015
Y2 - 18 May 2015 through 22 May 2015
ER -