How Search Query Can Be Optimized to Guarantee More Customer Satisfaction - Literature review Example

The performance of search engine is heavily dependent on the search query entered by the user. Currently, there is various search engines that provide search result with varied efficiency. This has prompted users to switch from one search engine to the other in search for more detailed and specific information. The four articles recommend how search query can be optimized to guarantee more customer satisfaction. In one of the article, collaborative ranking has been recommended for improving tail queries in search engine. On the second article, a system called metaphor has been analyzed in details to show how it can improve user experience by determining related queries and them offering relevant recommendation. The last article suggests the development of a query utility model that can assist in offering recommendation of queries with higher utility to users. In all the three articles, recommendations on how to assist the users to obtain the relevant information from search engines have been highly emphasized. This is due to the fact that the popular head queries do not wholly satisfy the users since they are deficient of specific and detailed information. How to address this problem in information search has therefore been addressed in details. Categories and Subject Descriptors H.3.3 [Information Systems]: Information Search and Retrieval General Terms Information, Design, Model Keywords Query Suggestions, Recommendation, Utility, Head Query Analysis, Collaborative Ranking, Relevance, Tail Query 1. Introduction In this paper, we analyze three articles which mainly revolve around improving the query search used by a user on the search engine. The articles are Reda etal [1], Zhou etal [2] and Zhu etal [3]. Each one of them suggests a different approach to solve a common problem experienced by users during information search. Therefore there exist some common themes that are evident in their approach. In the world today, there is an exponential increase in the demand for information which has been highly facilitated by the advancement of informational technology. This has resulted in the development of various types of search engines which help users obtain the required information successfully and in the shortest time possible. However, getting the relevant information efficiently has become elusive to many users prompting them to switch from one search engine to the other in search of more detailed and specific information [2]. In this paper, a summary of the way to optimize the informational search will be provided which largely try to address the problems faced by users globally. This is in regard to development and improvement of tail queries, determination of queries with higher utility for recommendation to users as well as highlighting the significance of metaphor in improving informational search in linkedln which has over 175 million users [1]. To better improve the search queries, user behavior is of much importance so that their experience while using the search engine can be statistically determined. In enhancing tail queries, one of the problems that emerge is data sparsity [2]. In addition, since tail queries are rarely submitted compared to head queries, the search for user behavior for proper collaborative ranking is made even more difficult. Modern search engine faces a common problem on how to help users formulate better queries for successful and satisfactory information search. A system such as Metaphor provides a platform which has helped millions of members in linkedln to optimize their search for other members for professional reasons since it recommends related searches [1]. Thus, the assistance of users in the informational search is paramount to their overall satisfaction. The general organization of this paper will adopt five sections to tackle different aspects in the three articles. In this regard, section two will tackle the common theme with the following section covering the discordant theme. Section four will dwell on the overlapping concepts which will involve the algorithms and concepts covered on each article. The last section will provide conclusion about the information provided in the three articles and how it can help improve user satisfaction while using a search engine to search for information. 2. Common themes In this section, we analyze the common themes between the three selected articles. In the three articles, it is clear that the main aim is to improve the significance of search results from search engines to ensure total customer satisfaction. All the articles propose different frameworks that ca help increase the utility of search queries. Their main aim is to improve query relevance and creating necessary recommendations. In the article [2] and [3], the authors try to suggest how a user can get the most relevant result from the search engine. In article [2] the authors propose the use of collaborative ranking to improve tail queries that are heavily affected by data sparsity. This due to the fact that there exist difficulties to determine the relevance of tail queries based on the inherent data sparsity problem. This is achieved through the use of the proposed CollRank (a collaborative ranking framework). The authors analyze those problems that affect tail queries and offer solution that can help address these problems. In using collaborative ranking, the authors basically suggest the utilization of data from multiple queries in a collaborative manner. This method assumes a joint supervised learning principle to determine the similarity measure between the ranking function and he queries. The measure of similarity is then use to find the relevance between queries and documents simultaneously. This proposed method improves the tail query relevance significantly. In article [3], the authors suggest the development of queries with high utility for recommendation to users. They try to address the tendency of users to failing to formulate a query that best describe their information requirements effectively. How to help those users thus becomes a problem for their search engine of choice. Developing a relevant query for recommendation is of much importance to help users get results quickly as well as manage to reformulate their queries to better suit their needs. They propose the use of query utility model o facilitate the capturing of query utility from users’ click behaviors and reformulation. In this way, the developed high utility query can help the users to accomplish their search activities successfully. In article [1], the authors’ suggest the creation of a unified data based from four different signals to develop recommendation to help members in linkedln conduct searches more successfully. They propose the use of filters and signals that capture the relatedness of searches from members. Theses are; collaborative filtering, query-result-query, overlap of queries (partial matching) and lastly the recommendation pipeline. In this way, the system metaphor can offer related search recommendation to the users. It is worth noting that the search of information is on the increase each and every day. The three articles highlight the problems faced by users during their interaction with search engine. They possess a common theme which seems to revolve around the optimization of informational search. The same has been successfully conducted in linkedln by use of a system referred as metaphor [1]. The successful optimization of informational search can however be achieved by improving the search queries. It is clear that head queries which are commonly used by the users fail to offer the required information leading to more time being consumed during research. Unfortunately, tail queries that provide the alternative are rarely used. To improve and optimize these tails queries, users’ behaviors have to be captured in advance [2]. Query recommendations help the user to reformulate their queries to better suit the information they are after. For users to gain from query recommendation, the suggested queries should be of high utility. The usefulness of queries necessary to better rank them is captured using the quality utility model [3]. However, mismatch of query information remains a problem hindering query ranking since it is hard to obtain the relevance of information extracted by use of tail query. Collaborative ranking used to enhance tail ranking doesn’t capture the information searched by different people. Thus, metaphor combines the collaborative filtering, query result query and partial overlapping recommendation in its design which translate to more clicks by users [1]. In this way, the user is assisted in refining their search by provision of alternate queries from the one created in their original search. 3. Discordant themes Though all the three articles suggest how to improve user experience on the search engine, how they go about it create the difference among them. One of them suggest the use of collaborative ranking to improve tail queries while the other suggest that ensuring that the query recommended are relevant, their utility has to be first determined. The third one proposes that use of relatedness of queries can help develop recommendations. It is thus clear that the three articles are discordant in the way they treat a common problem. The article [2] provides ways in which tail queries can be improved to guarantee more user satisfaction. In the article [3], it suggests the development of relevant queries while ensuring that they are of high quality. The article [1] highlights the efficiency of the system used to optimize search in linkedln which is wholly dependent on related search recommendation. 4. Non-overlapping concepts Since the selected article tackle a similar problem in different ways while ensuring that the overall result is to improve the search experience. a detailed analysis on each one of them would provide a glimpse of how they manage to influence the user experience while using the search engine. 4.1 Improving the Relevance for Tail Queries The article [2] proposes the use of Coll Rank as the collaborative ranking framework. Basically, it tries to use the similar queries in improving the ranking of tail queries. It involves computation of relevance score between he document and the tail query to obtain the ranking function. Then the scoring function is determined which represent the similarity between the first query and the second query. To learn the ranking function either square or pair-wise loss function is used. 4.2 High Utility Query Recommendation By Mining Users’ Search Behaviors In the article [3], the authors try to solve the major problem of learning the query utility with respect to users original information needs. To solve this problem they propose the use of Query Utility Model. To obtain the usefulness of the query, the users’ search sessions are investigated to deduce how he manages to get satisfactory results through a series of query reformulation. The utility of the query is further divided into perceived and posterior utility to represent the clicked result and the satisfaction from those results respectively. 4.3 A system fro related search recommendation The article [1] describes Metaphor a system used in Linkedln to recommend related searches to members. It combines four signals that enable it to develop a unified dataset that facilitate the generation of related search. One of the signal used s collaborative filtering which try to derive the similarity in click behaviors for queries. The second signal is query result query where other queries that facilitate a certain result to be clicked are identified to determine the click behavior. The third signal addresses the overlap in signals where unique queries are grouped and tokenization is carried out to enhance optimization. Then by use of empirical analysis, a unified recommendation dataset is created by combining the three signals. 5. Conclusion Information has increased over the decades which have been boosted by the advanced information technology. Due to the availability of large amount of information, the modern search engine has been hit by the problem of ensuring that user’s obtain the most relevant information in the least time possible. This problem can be addressed by the use of collaborative ranking to improve the relevance of tail queries which carry more specific information, or the assistance of user to reformulate relevant queries with higher utility. An example is the metaphor system used in linkedln which guarantee total user satisfaction in searching for new professional members. Coll Rank, Quality Utility Model and Metaphor system has been developed to help users in query formulation while interacting with search engine. In essence, they help in the creation or recommendation of queries with the highest degree of relevance to the user. In turn these can help them reformulate high utility queries or obtain alternative queries for them to derive highly relevant results. References [1] Reda, A., Park, Y., Tiwari, M., Posse, C., Shah, S., & 21st ACM International Conference on Information and Knowledge Management, CIKM 2012. Metaphor: A system for related search recommendations. Acm International Conference Proceeding Series, 664-673, 2012. [2] Zhou, K., Zha, H., Li, X., & 21st ACM International Conference on Information and Knowledge Management, CIKM 2012. Collaborative ranking: Improving the relevance for tail queries. Acm International Conference Proceeding Series, 1900-1904, 2012. [3] Zhu, X., Guo, J., Cheng, X., Lan, Y., & 21st ACM International Conference on Information and Knowledge Management, CIKM 2012. More than relevance: High utility query recommendation by mining users search behaviors. Acm International Conference Proceeding Series, 1814-1818, 2012. Read More