Query Enhancement and Query Recommendations - Case Study Example

Query Enhancement and Query Recommendations — This research is segregated in to three improvement areas: Firstly, the research is associatedwith executing queries for getting the desired and specific results from the search engine. Secondly, the research discusses about the recommendations of news associated with the items, as it will facilitate the user to take a decision of buying any product online. Thirdly, the research is about children who struggle to get the desired results from the web search due to their vocabulary limitations and difficulty in selecting keywords. The research recommends URI presence and meaningfulness methods for establishing accuracy in executing queries. Moreover, a news recommender mechanism is discussed for assisting online shopping and query suggestion mechanism, shared web links and social media tags are utilized for assisting children to the desired search engine keywords .Index Terms – NAQ’s, URI, Query I. Overview The research describes a detailed characterization model based on the natural queries that are representing real life scenarios along with a comparison to deploy world most popular search engine mechanisms. Moreover, the research also incorporates an addition of an explanatory statement for facilitating users more towards the items they like and why an item needs consideration. Furthermore, a research on an engineering of textual entity based explanation is also discussed, as Markov logic networks are integrated to rank user reviews via tags taken from social networking sites, as they enhance browsing experience for children. In addition, we have also discussed query and search engine based query suggestions and random walks. II. Common Theme An uncommon keyword typed in the search bar end up with a very few results and sometimes with no results at all. For instance, some uncommon file named generated by a malicious code or virus. The keyword is usually long in in terms of characters or having a different language (Blanco, Ceccarelli, Lucchese, Perego, & Silvestri, 2012). Likewise, the search engine is capable of handling such queries via different mechanisms. The research will explore the queries that are based on the numbers or query those results in no matches. (Blanco, Ceccarelli, Lucchese, Perego, & Silvestri, 2012). Moreover, the study also focuses on several ways of developing explanations for the suggested news, for this reason, a method embedded with machine learning is introduced for enhancing the usefulness of the news recommendations (Altingovde et al., 2012). Likewise, a black box is developed that is also called a news recommender. By exploring the news recommender, few risks were identified such as automated new recommendations are itself a new issue. Previously, study shows that it has focused on dissimilar visualization and media recommendation strategies (Altingovde et al., 2012). Moreover, the study also demonstrated query suggestion mechanism to facilitate children for keywords that are useful for them. However, research has been conducted on query suggestions and reflects a deviation i.e. the suggestion query focuses on search intents that are potential for the children, the construction of these suggestions do not maintains any query logs and the ranking of these suggestions relies on random walk for promoting focused on topic of interest for children. The suggestion of these queries relied on the mechanism that is based on the social networking tags extracted from the bookmarking mechanism. ‘Delicious’ is the work that is linked with the query results from the web, as previously it was recognized as web links proposed for children (Duarte Torres, Hiemstra, Weber, & Serdyukov, 2012). Likewise, ‘Delicious’ is considered to be a valued tag in the domain, as the queries submitted by the end users can be exploited. Moreover, the collective information submitted by the end users can be forwarded via shared web links for children or resources (Duarte Torres, Hiemstra, Weber, & Serdyukov, 2012). III. Discordant Themes The queries that we are now highlighting are based on the question that repossess no answers (NAQs). Moreover, in an individual search engines, we are now focusing on a specific questions that have no answer (Altingovde et al., 2012) In order To inspect about the description and features of the questions we need to carry out a research. In addition, this research is based on the study performed on the users. This research is judge by the four researchers and the NAQ is mentioned on the basis of two kinds of test (Altingovde et al., 2012): URI presence Meaningfulness The initial test that is URI presence is carried out to inquire about the presence of URI in NAQ. In order to make this URI presence test automatic then the pattern matching technique is utilized. However, the best results can be found by performing this test manually. Another reason to conduct the test manually is that is difficult to find the URIs that includes typos the results in Fig. 3. The research indicates that at least one URI is present in around 57% of the NAQs. Moreover, distorted URI are present at the percentage of 45%. The remaining URI that are present in NAQS are normal and are estimated at 55%. The results revealed that around one-third of NAQs are targeted to recover assets that are not known or are not searchable by the search engines. (Altingovde et al., 2012) Thus, it is proved that the questions regarding NAQs are not an unsolvable issue and can be resolve by any method. The three search engines named as A, B and C has taken part in this research. The result concludes that the search engine A has performed well by resolving questions related to NAQs that also includes distorted URIs. In The amount of NAQs present as compared with the three search engines are little high in search engine A. in general, the union size is comparatively low as compared to the size of the intersection in most NAQs. These NAQs also includes URI and are solved by the individual search engine (Altingovde et al., 2012). Another test that is performed by our researchers is based on the meaningfulness of the NAQs. In this test we will only believe the URIs that are left behind in a question that includes URI. In addition, if the whole NAQ includes URIs then it is excluded from the test and named as “only URI”. Moreover, the NAQ has the ability to give meaning to its users if the judge does not understand the meaning of NAQ (Altingovde et al., 2012). In fact, the NAQ has a power to ensure its meaning and then the judge will mentioned the NAQ as “unsure”. There are some NAQs that are not clear to the judges and labelled as meaningless such as, queries that are only formed of repetitive key strokes are mentioned as “meaningless”. Furthermore, the left over NAQs are then mentioned as “meaningful” and “no typos” present. (Altingovde et al., 2012) There is a large amount of NAQs that are mentioned as |unsure” therefore, for the lower bound the NAQs are utilized that are remaining. (Altingovde et al., 2012)The results reveal that there are only 3% of NAQs present that are considered meaningful and does not include any typos. In this research, we only come across an individual NAQ that is not resolved by any of the three search engines. A typo is present in At least four out of five each NAQ that is meaningful. However, this typo in NAQ cannot be set by the spell checker. In fact, the NAQs that do not have any meaning are At least 21%. On the portion of NAQs, the concluding result shows an upper bound of 79% that can resolve by a search engine through an easy technique (Altingovde et al., 2012). II. non overlapping concepts In recent times there is always a challenge to find out a news commendation system in order to justify the models for research in a right way. Particularly, the explanations as conducted by the news system do not target the elevated ratio. In fact, the user’s interest is developed in the news on condition along with extra information on the suggestions that are being proposed. The core task of news system is to generate the user’s attention. This can be done by providing genuine news along with added information over the news platforms. These also include the right explanation regarding the recommendation that has been publicized. The queries that are not answered on the famous Yahoo! web search API are recovered by another search engine. In this research our policy might look slightly partial in terms of those queries that are not resolved by the Yahoo!. In order to examine this matter, a random pick is made of 6 thousand singleton queries that are derived from the AOL query log. These 6 thousand queries are not from our initial bundle of queries of 16 thousand. In addition, the non-singleton queries are defined as those queries that can be resolved by all three search engines. The queries are mentioned on all the search engines and the results are concluded. The results reveal that the position of search engines according to the percentage of NAQs is mostly identical as Mentioned in Table 1. On the other hand, the result shows that the numbers are little low as expected. However, the research that is conducted shows the results that are not manipulated against any of the three search engines. Nonetheless, the results must be authenticated with help of the seed sets that are focused on the search engine’s APIs. This could result in a positive research direction for future testing. Moreover, during this research this should also be put under consideration that the results that are concluded from yahoo web are provided by the Bing. However, our preference would be two search engines that took part in research. In the new form there are a number of queries that are tough and could barely go the search results. Conversely, the search engines are certainly very potential in order to resolve a number of tough queries. Moreover, the performance is also enhanced with the help of significant mechanism of query correction. This is based on the number of resolved queries that are mentioned in the results. For the large number of queries that are tough to resolve are also corrected by this mechanism. The correction mechanisms has common technique in order to resolve connected query and work along with the proposal patterns that emerge in search result pages. The search engine is made to handle all kinds of typos that are present in NAQ. The correction of typos can be done with or without consideration of their number. The qualified analysis based on the research results reveals that the search engines are always keen to resolve the typos and the tough queries through a simple technique. The research results also show that how the tough queries are managed and solved successfully by the correction system. The study for the result page that is been rediscovered shows the result patterns that are attempted by the three search engines. The result patterns are four in number. Initial pattern shows the answer related to the accepted query. The second pattern shows the results that are provided by the three search engines. The result included the return form of query correction/implication along with the results of the new query. The pattern that is observed in the third query shows the results related to other the query instead of the submitted query. Lastly, fourth pattern shows no results related to the submitted query. III. summary and conclusion This research discusses a complete mechanism of characterization based on the natural query logs representing real life and a comparison of the deployed mechanism by world’s most popular search engines. Although, there is always a room for improvement, as few queries are labeled as no answers query i.e. NAQ. The intensive research showed that some queries are incorporated with URI and the developed mechanism can be accomplished for resolving 79% of NAQ aimed at URIs. Moreover, the paper also discusses about an addition of an explanatory statement for facilitating users more towards the items they like and why an item needs consideration. Furthermore, a research of an engineering of textual and entity based explanations and utilizing Markov Logic networks to rank user reviews is also discussed with the utilization of tagging extracted from social media for enhancing web browsing searching experience for children. Lastly, we have also discussed query and search engine based query suggestions and random walks. References Altingovde, I. S., Blanco, R., Cambazoglu, B. B., Ozcan, R., Sarigil, E., & Ulusoy, \. (2012). Characterizing web search queries that match very few or no results. Proceedings of the 21st ACM International Conference on Information and Knowledge Management, Maui, Hawaii, USA. 2000-2004. doi:10.1145/2396761.2398560 Blanco, R., Ceccarelli, D., Lucchese, C., Perego, R., & Silvestri, F. (2012). You should read this! let me explain you why: Explaining news recommendations to users. Proceedings of the 21st ACM International Conference on Information and Knowledge Management, Maui, Hawaii, USA. 1995-1999. doi:10.1145/2396761.2398559 Duarte Torres, S., Hiemstra, D., Weber, I., & Serdyukov, P. (2012). Query recommendation for children. Proceedings of the 21st ACM International Conference on Information and Knowledge Management, Maui, Hawaii, USA. 2010-2014. doi:10.1145/2396761.2398562 Read More