Revisiting Contextual Recommendation from an Information Retrieval Standpoint
Citation:
Chakraborty, Anirban, Revisiting Contextual Recommendation from an Information Retrieval Standpoint, Trinity College Dublin.School of Computer Science & Statistics, 2021Download Item:
PhD_Thesis_Anirban.pdf (PhD Thesis) 4.661Mb
Abstract:
The challenge of providing personalized and contextually appropriate recommendations to a user is faced in a range of use-cases, e.g. recommendations for movies, places to visit, articles to read etc. In this thesis, we focus on one such application, namely that of suggesting 'points of interest' (POIs) to a user given her current context(s), by leveraging relevant information from her past preferences. An automated contextual recommendation algorithm is likely to work well if it can extract information from the preference history of a user (exploitation) and effectively combine it with information from the user's current context (exploration) to predict a POI's 'appropriateness' in the current context. To balance this trade-off between exploitation and exploration, we propose a generic unsupervised framework involving a factored relevance model (FRLM), comprising two distinct components, one corresponding to the historical information from past contexts, and the other pertaining to the information from the current context.
We further generalize the proposed model FRLM by incorporating the semantic relationships between terms in POI descriptors with the help of kernel density estimation (KDE) on embedded word vectors. Additionally, we show that trip-qualifiers, such as trip-type (e.g. vacation, work etc.) and accompanied-by (e.g. solo, friends, family etc.) are potentially useful sources of information that could be used to improve the effectiveness of POI recommendation in a current context (with a given set of these constraints). Using such information is not straight forward since users' text reviews of POIs visited in the past typically do not explicitly contain such annotations (e.g. a positive review about a pub visit does not contain information on whether the user was with friends or alone, on a business trip or vacation). We propose to use a small set of manually compiled knowledge resources to predict the associations between the review texts in a user profile and the likely trip contexts. Our experiments, conducted on the TREC contextual suggestion (TREC-CS) 2016 dataset, demonstrate that both factorization and KDE-based generalizations of the relevance model contribute to increased effectiveness of POI recommendation. Further, we demonstrate that trip-qualifier enriched contexts further improve the effectiveness of our proposed model.
As we explore IR-based approaches (specifically pseudo-relevance feedback methods) for contextual recommendation, we also seek to estimate a robust set of feedback documents by, generally speaking, employing a document selector function to decide which documents are useful in improving the quality of relevance feedback. To mitigate the problem of over-dependence of a pseudo-relevance feedback algorithm on the top-M document set, we make use of a set of equivalence classes of queries rather than one single query. These query equivalents are automatically constructed either from a) a knowledge base of prior distributions of terms with respect to the given query terms, or b) iteratively generated from a relevance model of term distributions in the absence of such priors. These query variants are then used to estimate the retrievability of each document with the hypothesis that documents that are more likely to be retrieved at top-ranks for a larger number of these query variants are more likely to be effective for relevance feedback. Results of our experiments show that our proposed method is able to achieve substantially better precision at top-ranks (e.g. higher nDCG@5 and P@5 values) for ad-hoc IR and points-of-interest (POI) recommendation tasks. Primary motivation of this part is to achieve better precision at top-ranks by improving the quality of relevance feedback for IR in general, which is eventually applied in the specific task of contextual POI recommendation. POI recommendation, being a precision-oriented task, provides an interesting use-case to study the robustness effects of relevance feedback.
Sponsor
Grant Number
ADAPT Centre (Grant 13/RC/2106)
Author's Homepage:
https://tcdlocalportal.tcd.ie/pls/EnterApex/f?p=800:71:0::::P71_USERNAME:ACHAKRABDescription:
APPROVED
Author: Chakraborty, Anirban
Advisor:
Conlan, OwenPublisher:
Trinity College Dublin. School of Computer Science & Statistics. Discipline of Computer ScienceType of material:
ThesisCollections:
Availability:
Full text availableLicences: