International Journal of Engineering Science Invention

ISSN (Online): 2319 6734, ISSN (Print): 2319 6726

www.ijesi.org Volume 3 Issue 4 April 2014 PP.21-25
www.ijesi.org 21 | Page
Mobile Search
1,
Amol D Gaikwad ,
2,
Bhakti S Kurhade


It ,Ycce Rtmnu, India
Cse, Ghraet Rtmnu, India

ABSTRACT : A UPL is an indispensable source of knowledge which can be exploited by intelligent systems
for query recommandations, personalized search, and web search result ranking etc. This paper focuses on the
personalization strategies which explicitly and implicitlyinfer user search context at individual user level. Web
service is a popular standard to publish services for users. However, diversified users need to access web
service according to their particular preferences. Mobile search is quite different from standard PC-based web
search in a number of ways: (a) the user interfaces and I/O are limited by screen real state, (b) key pads are
tiny and inconvenient for use,(c) limited bandwidth and (d) costly connection fees. These limitations result in
more navigational queries in the mobile search. Mobile search is becoming increasingly important for mobile
users as mobile devices are more widely used. Most importantly, user ontology with both static and dynamic
information is defined by web ontology language (OWL), through which the location server can do Personalized
search of servicesWe propose an architecture which collects user information(at mobile device and carrier
network) and derives user intention in given situations. We show that personalized mobile search perform well
for ambiguous queries.
KEYWORDS: Mobile Search, Personalized Search, Context Awareness, User Profile

I. INTRODUCTION
Researchers have explored personalized search to improve topical relevance of result documents in PC
based web search. The users immediate and short-term search context to expand the current query mechanism
based on the user interest. Chirita et al. [9] proposed personalized search and summarization algorithms which
assist based on extracted information from local desktop. Mobile search is the second most used application
only after social networking in wireless internet .Search engine like Google appears in top three of the most
visited web sites in terms of wireless internet usage. Most mobile are short due to the hardware limitations such
as tiny keypads and small screen. The top 100 mobile queries at AT&T [3] reveal that a great number of search
queries are navigational [4] in nature. The navigational searches, for example Google and Bing, usually steer
mobile users to Specific web sites conveniently. Unlike navigational queries, words like software and free
that are informational and transactional are ambiguous to search engine. A housewife and an iPhone user
interpret apple differently in search context. A housewife is likely to know the apple variety and prices at the
local grocery stores. While an iPhone user is interested in service or products related to iPhone. A variety of
contextual information is acquired through various sensors such as a GPS device, temperature, noise, and air
pollution and interacting with other application services such as weather and traffic services based on Web
services technology. The data transmitted from other sensors, with the exception of GPS sensors, can not be
generally used in context aware systems without the installation of sensors and construction of wireless sensor
networks even though the data provide more contextual information. For this reason, most approaches of
context-aware systems have service-oriented architecture based on Web services with context information We
propose a smart proxy on the wireless device to derive user context profile. A dynamic neural network based
algorithm is utilized to the users given situation, location, activities, and history informationThe introduction of
the paper should explain the nature of the problem, previous work, purpose, and the contribution of the paper.
The contents of each section may be provided to understand easily about the paper.

II. ARCHITECTURE DESIGN
The mobile devices play an extensive role of collecting, analyzing, and extracting context entities.
Context profiles compiled at client side adapt to the mobile users current situation.
Mobile Search
www.ijesi.org 22 | Page

Figure 1: Proposed System architecture
Figure 1 shows the paradigm of the client/server model. The user inputs (voice or data) and surrounding
environment inputs (temperature, GPS reading, altitude,e.g.) are collected by the hardware logic or the
applications such as the user calendar. The traditional client/server paradigm fits well into the mobile search
application. The carriers network is the server which provides data and voice services to subscribers. User
submits mobile queries to network. The context aware proxy further inspects inputs and extracts context entities
from the inputs. Finally, the proxy compiles the context profile and sends it to the carrier server.The carrier
server learns the user situation from the context profile in addition to data collected at the network. For example,
the network context unit collects user information, i.e., location and query history. The intelligent manager
carries out a few important functions such as identifying the ambiguous queries and expanding these queries
with context profiles. Expanded query contains extra information which could facilitate search engines and
improve the topic relevance of returned documents. duo et al. [10] and Teevan et al. [11] investigated the
personalized search strategies and stated that personalization improves the search accuracy on ambiguous
queries. So far the personalization is studied only for PC based web search. Most of such personalization
strategies are limited to the user search history, returned search results, and documents stored in PC. We extend
this line of work into mobile search and derive user context based on profiles which adapt to user location,
activities, interaction history, and preferences. We also believe that the increased usage of location-based
applications and services will lead mobile users to search for local services and information. Liu and Birnbaum
[12] developed Local Survey which aggregates local views associated with news events or topics. Vadrevu et al.
[13] pointed out the importance of identifying the regional sensitive queries. Hence, the mobile search must
cope with local search query .We propose a proxy on the wireless device to derive user context profile
III. MANAGER
One of the trends in the wireless device is to integrate more sensors, e.g. tilt sensor and touch screen
LCD on iPhone. The software applications on wireless devices advance in parallel. Hence, context proxy on
wireless devices is able to collect user information, analyze user search context, and compile context profiles.
3.1. Profiles
The profile [14] is a collection of context entities extracted from the on-board sensors, client
applications, user activities, and so on. Ideally, context profiles should be a good reference of users current
situation from which the carriers network could derive users context or intention. Thus, the proxy on mobile
devices frequently updates context profiles and uploads them to network for reference if necessary. Due to the
limited bandwidth [15], the context profiles should be concise. As user changes activities/context, context
entities are added to or dropped from the profiles based on the algorithms we specify in the later section.
Whenever addition or deletion of context entities occurs, the proxy notifies network the changes. sent to
network. There are four context profiles managed by the client proxy;

[1] User profile describes the users habits and emotions In addition, this profile includes the software
applications and user interfaces. Examples of context entities in this profile are processor, memory, display
type, device size, and OS.
[2] Profile describes of context entities in this profile are processor, memory, display type, device size, and OS.
Mobile Search
www.ijesi.org 23 | Page
[3] Environment profile describes surrounding area of the mobile user. This profile stores user location.
3.2. ANN Algorithm
Artificial Neural Networks (ANN) are widely utilized in many of today's novel applications for solving
practical problems, such as pattern recognition, classification, forecast studies, and cluster analysis. The learning
rule is based on the following general equation:
Wij (t) = xi(t)yj(t) (1)
We modify xi(t) as a function of the context changes DC and frequency count F of the context entitiesin the
context cache managed by the client proxy between the current epoch and the previous epoch. In addition, yj(t)
is defined as the weight values, reference W , in the current user situation.
Thus, the derived equation of weight change for our approach at epoch t follows:
W (t) C(t) F(t) ( ) k , W t |k k reference = + ( 2)
where DC(t) is calculated ascurrent user situation., the derived equation of weight change for our
approach at epoch t follows:W (t)=| C(t)+m F (t) | k , W k k reference
where DC(t) is calculated as

C( t) = |C( t)- C (t-1)|
C( t -1)
C(t) is the current context value, and C(t -1) is the previous context value. If DC(t) is zero, then
C(t) = -|C(t) C average| .
C average is the mean of C(1) through C(t-1). F(t) is the difference in n frequency counter for contextentity
between at current epoch and previous epoch, F(t) = F(t) - F(t -1) .
is the coefficient that is machine specific. Finally, the weights for context entity k at epoch tare updated as
Wk(t) = Wk(t-1) + Wk(t) (3)
If the current user location contributes new context t weights that do not exist in the previous context setting,
then the initial value of Wk(t) is set to 1.Environment profiles, user profiles, and data profiles are updated based
on the above equations such that these profiles adapt to current user situation. Ambiguous queries submitted to
the search engines are expanded with the more user information from these profiles.

IV. MANAGER
Manager is the most important component at the carriers server side. It learns the mobile users
context information at the network level and analyzes the next profiles sent from the client devices. When
mobile user submits query manager indentifies the ambiguous queries and further expands these queries with
derived user context. The expanded queries are then submitted to search engines
4.1 Search Query Expansion
The context profiles compiled by proxy at client devices reflect the hardware configuration, user
applications, usage history, and derived user situation.Higher weights mark the importance of context entities.
Of course, not all context entities in the profiles have changing weights. The hardware profile has weight fixed
context weights due to the stable hardware and software configuration at the client device. Context profiles
include the weights of the context entities calculated at proxy, p = [CWi+1, CWi+2, CWi+3, , C Wi + n],
where i _ [0, 1, 2, 3]. Let the query forms the set, q = [T1, T2, T3, , ]. After query expansion, the
augmented context profile is
m n
AP = CW i + K ( Tj) (4)
i=1 i=1

where K(Tj) is the offset function. An example augmented context profile could be AP = CWi+1,CWi+2+
K(T2), CWi+3, , CWi+n, K(Ti+n+1), ]. If the terms/words from the query set are matched in the context
profile, then the offset weight of the term/word,K(Tj), is added to the context weight, CWi+n. Furthermore, if
the terms/words from the query set are not found in the context profile, then the offset weights profile. It is
obvious that the augmented context profile is likely to have more weights on the query words.


Mobile Search
www.ijesi.org 24 | Page
4.2. Documents Ranking
The search engines return all the documents related to the query based on the augmented context
profile. The returned documents are grouped into categories and ranked in terms of descending weights in the
context profile. Categories that match more context weights in augmented context profile have higher rank.
Because most mobile devices have small displays, the maximum documents in each category are limit to a
number. This method relies on the user selection to further narrow down user current context. When the user
selects the data or services downloaded from the server, his/her selection would greatly help the system
understand users intention..

V. EXPERIMENT
In this scenario, the user carrying. During this he submits search query. appleThe returned
documents show that the personalized search results include documents related to the mobile query. Key word
apple is ambiguous without specifying location. Because the location is included in the user context, then the
search engine understands it is a search in area. In addition, the context profiles reveal another important topic
which is food. Thus, the returned documents also include links related to food or restaurants in the local area.
The context profiles assist the search engine to provide local information selection on the data allows the related
context weights to feed back into the proxy. This feedback mechanism significantly improves the accuracy of
the server response.


VI. CONCLUSION
With the wide spread of mobile devices, mobile search becomes an important application for mobile
experience. Mobility, social activities, and locality add to the sophistication of mobile search. Personalized
search has been the focused approach for PC-based search lately. We extend this line of work to mobile search.
We propose a client-side context proxy which collects user information, analyses context information, and
compiles context profiles. These context profiles, which reflect user current context, assist query expansion to
solve ambiguity. An intelligent manager is proposed to process context profiles, identify ambiguous query,
expand query, and reorder the result documents based on user current topics in the context. The simulation result
shows that the proposed architecture can adapt to user situation and provide local information for mobile search.
REFERENCES
[1] Opera Software, State of the Mobile Web Report: First Quarter, May 20, 2008
[2] A. Soffer, Y. Maarek, and B.W. Chang, WWW2002 Workshop on Mobile Search, Mobile Search, Honolulu,Hawaii, May,
2002.
[3] B Meunier, Characteristics of the Top 100 Mobile Search Queries at AT&T, Jan, 2008.
[4] B.J. Jansen, D.L. Booth, and A. Spink, Determining the User Intent of Web Search Engine Queries, The International World
Wide Web Conference, Banff, Canada,May 8-12, 2007.
[5] S. Cronen-Townsend and W.B. Croft, Quantifying Query Ambiguity, Proceedings of the Conference on Human Language
Technology, San Diego, 2002, pp. 9498.
[6] R Song, Z.X. Luo, J.R. Wen, and H.W. Hon, Identifying Ambiguous Queries in Web Search, The International World World
Wide Web Conference, Banff, Canada, May 8- 12, 2007.
[7] X.H. Shen, B. Tan, and C.X. Zhai, Implicit User Modeling for Personalized Search, Conference on Information and
Knowledge Management, Bremen, Germany,2005.
[8] F. Qiu and J.H. Cho, Automatic Identification of User Interest for Personalized Search, The International World Wide Web
Conference, Edinburgh, UK, May 22-26, 2006.
[9] P.A. Chirita, C.S. Firan, and W.Nejdl, Summarizing Local Context to Personalize Global Web Serach,
Information and Knowledge Management, Arlington,Virginia, November 5-11, 2006.
Mobile Search
www.ijesi.org 25 | Page
[10] Z.C. Dou, R.H. Song, and J.R. Wen, A Large-scale Evaluation and Analysis of Personalized Search Strategies,International
World Wide Web Conference, Banff, Canada, May 8-12, 2007.
[11] J. Teevan, S.T. Dumais, and D.J. Liebling, To Personalize or Not to Personalize: Modeling Queries with Variation in User
Intent, ACM SIGIR Conference,Singapore, July 20-24, 2008.
[12] J. Liu, and L. Birnbaum, What do they think? Aggregating Local Views about News Events and Topics,International World
Wide Web Conference,Beijing, China,April 21-25, 2008.
[13] S. Vadrevu, Y. Zhang, B. Tseng, G. Sun, and X. Li,Identifying Regional Sensitive Queries in Web Search,The International
World Wide Web Conference, Beijing,China, April 21-25, 2008.
[14] F. Gui, C.M. Liu, X. Chen, and D.D. Wang, A Context-Awareness Algorithm Based on Data Cache for Mobile Computing,
World Wireless Congress, San
Francisco,May25-28,2004