Week 4: Users & Information Seeking

LIS 4/5523: Online Information Retrieval

Dr. Manika Lamba

Introduction

This week, we’re going to talk about users and information seeking behavior. This becomes very important to us: one, as someone who’s helping users find information in retrieval systems, but also for those that design information systems and databases that help users connect with information.

Important Questions

1. Who are Users?

2. What can we know about Users?

3. How do users seek and search for information?

   Examples of models for search and information seeking behavior

So let’s talk about our users. Some questions to frame our thinking:

• Who are our users?
• What can we know about users?
• How do users seek and search for information?

We will also look at some different models that have been developed in Library and Information Science [LIS] through the years, primarily starting in the 1980s about how users look for information. So, these are some questions to help frame our thinking. Let’s talk about each of these aspects individually.

Who Are Users?

A user is a person who uses information and/or information systems in some meaningful way

First, let’s look at “Who are our users?”

A user, very broadly defined, is “a person who uses information and/ or information systems in some meaningful way.”

Now, meaningful is a very subjective term, and you’re thinking “what does she mean by meaningful?”

And generally, it’s up to the user to decide if their interaction with the system has been useful.

Now, I’m not just talking about automated systems, technical systems. I’m talking about systems in general.

You know, walking into the library and talking to a reference librarian would be part of a system’s interaction. I’ll explain that in just a few minutes, but users are anyone that really uses our system to find what they’re looking for, to solve an information need. And there are lots of different terms that you’re going to see in the literature that refer to users.

In the 80’s and the 90’s we’ve called them “patrons” and “end users” and “system users”. We used to refer to librarians, specifically reference librarians, as search intermediaries, so you might see that in the literature, but that was primarily in the late 80’s and early 90’s. But now we just call them users.

Today’s User

[Click Next to view the pictures]

Okay, so who exactly is today’s user?

I’ve given you some pictures on this slide to think through what a user or who a user is, but one thing that you’ll notice, of course, is that there is either a computerized or virtual or a print based technology in place in each of these pictures.

You will notice, computers, much of what you would see in most of the libraries today with various different people of different genders or ages accessing the Internet or using computer systems within their library context.

You also have to think about users in terms of age groups. We have young users that are interacting with and adapting technology in ways that we did not anticipate, which is very exciting for us as library and information professionals, but it’s also very challenging because we now have to think about how do we engage with the users and help them become better information consumers as well as creators of information.

And we cannot forget that people still like to browse our libraries. People traditionally are used to going through the stacks, finding something they like and then looking at the books around the book to determine if there’s something else that they find of interest or that’s related to their information need. So, classification systems are still very relevant in libraries. That’s not to say that the systems in place are the best systems or the most user-friendly systems for this new generation of user, but they do still support users in very important ways.

A really interesting dimension of users, and this came about, of course, with the Internet and more social ways of sharing information or participating in online communities, is what is being called “citizen science”.

Also, we have very wired individuals as users today where a user is connected to internet 24/7. These users are using different tools like smartphones or tablets in ways that allow them to access information but also to share and create information and engage in different practices with other users. These are all dimensions of users we as information professionals have to think about.

Model of Today’s User

Work in Progress?!

So, when we think about today’s users, it’s really difficult to model the user today because of all the different technology that they interact with on a daily basis and all the different resources they have available to them at their fingertips.

So, let’s think about the user as a “work in progress” as we work through this lecture on users and information behavior.

Some More Vocabulary

• Information: any difference you perceive in your environment or within yourself
• Information need: recognition that your knowledge is inadequate to satisfy a goal you have
• Information seeking: conscious effort to acquire information in response to a need or gap in your knowledge
• Information behavior: information seeking, as well as other unintentional or passive behaviors, or other purposive behaviors such as avoiding information

Some other vocabulary that might be useful for understanding this dimension of information seeking and organization of information.

In this program you will read lots of definitions of information. One definition I find particularly useful is: information is any difference you perceive in your environment or within yourself.

So, we can think of information in lots of different ways. We can think of it as a cognitive change that comes about when we learn something new or we encounter some new piece of information. We can think of “information-as-thing,” or as physical objects actually being thought of as information bearing Or having the ability to inform us in some way. So, there are lots of different ways we can think of information.

Information need then is that recognition that you have that your knowledge is somewhat inadequate in a particular area and you need to find something else out in order to satisfy a goal that you have. For example, if you want to learn about designing webpages, you’re probably going to go to a class, or you’re going to find a book that will help you with this, or you might even go on the web and find different information. But you’re going to access systems, resources, in order to find information to solve your information problem.

Another term is information seeking, and this is a conscious effort to acquire information in response to a need or gap in your knowledge. Think about your daily activities. Many of you probably got up this morning, turned on a computer or tablet, turned on the television, listened to the news, got information about the weather, so you could decide what you’re going to wear today or if you need an umbrella. You probably have talked on the phone to someone today. At different points during your day, you have been encountering or accessing different systems to find information.

Information behavior, then, is kind of this overall umbrella of information seeking as well as other unintentional or passive behaviors or other purposeful behaviors related to seeking (or not seeking) information, such as avoiding information.

Now, how many of you have heard of cognitive dissonance? It comes from cognitive science. Cognitive dissonance is where you avoid information that you know is going to change something that you’ve already made a decision about or something that you’ve already done and you don’t really want to encounter that information because you know it’s going to make you feel bad about your decision. That’s just one way of looking at it. So, for example, you’ve just made a major purchase, like you bought a new car. And then you get on the website or you hear consumer reports or something telling you how bad that car is and that you really shouldn’t have bought it in the first place, and that’s the idea of avoiding information.

Also, it’s interesting in our field because of what we call the information glut, or the information overload. That people purposely filter our information sources in their lives because they don’t want to be overwhelmed with all this different kinds of information. For instance, I don’t turn on the news before 8:00 because I don’t want to know what’s going on until I’m ready for it.

But that’s one way in which we avoid information. Turning off the radio while you’re driving, things like that. When we look at information behavior, we look at not just purposeful information seeking, but we look at avoidance activities as well.

So, it’s a really interesting area of our field that tells us a lot about people and how our information can be represented for them in order to effect access to that information.

So, what do we need to know about the people we call our users?

What do We Need to Know?

• Demographics
• Types & Levels of Knowledge
• Information-Seeking Situations

Well, there are certain things we can know, and there are certain things we can make some generalizations about, such as demographics, types and levels of knowledge, and information seeking situations. Let’s address each of these separately.

Demographics

• Age
• Gender
• Education
• Occupation
• Culture (values, language, ethnicity)

Age is one way that we can get a good picture about a user group. And we can characterize age in many different ways. Our society has created these frameworks in order for us to do so. We can think about age in terms of K-5, K-12, young adult, adult, senior citizen, children. We put all these different labels on it. So, that’s one way in which we can get a handle on one characteristic of our user group.

Also, gender, are they primarily males and females? In earlier studies of computer adoption studies this demographic was a real factor. It’s not as much today as it used to be, but again there are still studies that show that males are heavier users of different kinds of information technology than females are, but that’s shifting.

What’s really interesting about some of our studies today with gender is that we’re seeing that particularly in the web and in the internet environment is that we have different uses – that each gender has a different use. Females tend to use more of the communicative aspects of the web and internet than males. And males generally – and I don’t like to generalize, but studies have shown this – are more into the information gathering and gaming and more of the high tech applications.

So it’s kind of interesting what we can tell about a user population based on the different variables. But this is shifting constantly.

Education can tell us a lot about a user group. It tells us about their potential knowledge, about their subject and domain expertise; it tells us about potential uses they might have for that information that they’re looking for. So, education level can tell us quite a bit.

Occupation is the same. If we know that someone is an engineer or chemist or a contractor or a builder, they’re going to have a different knowledge base potentially. And so it’s interesting to be able to tell what their occupation is because that again will tell us something again about what they’re going to be looking for, what kind of resources would be useful for them.

Culture is a little bit of a grayer area. But it does tell us a little bit about values and ethnicity and ethnic choices that people make. But also language. Especially if you’re designing a system. You need to know what kind of language or what languages your users need in order to access your system. Is it a multilingual system? Do you have a very heavy multilingual population, and how do we adapt that within the system environment? So, even language can be a very important variable to know about our users. We have a sparsity of studies related to cultural aspects of information behavior and users.

Types and Levels of Knowledge

Allen (1991)

• World Knowledge
• System Knowledge
• Task Knowledge
• Domain Knowledge

The second set of information that we can know about our users is what we call “types and levels of knowledge”:

• world knowledge
• system knowledge
• task knowledge
• domain knowledge

And others have expanded on this idea as well. So, let’s break them each down.

1. World Knowledge

• World View, Perception of Self

• Affects Self-Confidence

Now, we can look at them separately, but you can also see the connections and the overlap between all these types of levels of knowledge.

For example, world knowledge is a user’s perception of themself and the role that they play in the world – in their own world. And it’s really a combination of the other three knowledge rolled into one. This is kind of from a constructivist learning point of view – that you build or construct your own world based upon what you encounter, what’s in your society, and that frames and shapes your perception of yourself and the role that you play.

Why this is important to us is that it will affect a user’s self-confidence when they’re working with a particular system, when they’re looking for information, when they’re coming to the reference desk and asking questions. So, somebody with high world knowledge, someone who has traveled a lot, someone who has a real broad view of the world and different pieces of the world, and has a good sense of how they fit in the world, is probably a more confident user of information and information sources.

2. System Knowledge

• Understanding of Information Environment

Information systems at some level (e.g. cultural, institutional, technical, interpersonal)

• Affects Skills and Expections

The second type of knowledge is a little bit more complex. This is system knowledge, and this has been expanded upon by many people in our field.

For example, Allen defines it as the understanding a user has of an information environment. And this can mean many different things. This can be an information system at any level – not just a technical level. It can be a cultural level – how their community system works. Who do you call if you need to find out if they’re picking up your garbage on a holiday? That’s an example of your cultural system. An institutional system would be something you know, “Who do I call if someone’s put a lock on my car wheel, so I can go home?” But you have to understand how that institutional system might work.

Technical is related to knowledge of how the technical system works, such as, “How does this database work? What can I do to make my search effective?”

Interpersonal systems are your personal network. Who do you call if you need a recipe for coffee cake? So, your own personal network, your friends, your family, your colleagues.

Your understanding of system knowledge affects your skills and your expectations when you’re using different systems.

Okay, now let’s expand this even further.

2. System Knowledge (Cont.)

• Conceptual knowledge of the information retrieval process - - translating a need into a query

• Semantic knowledge of how to implement a query in a given system - - the how and when to use the system features

• Technical skills in executing the query (Borgman, 1996)

  • basic computing skills
  • syntatic knowledge of query entry and how search statements, systems work

System knowledge includes many different aspects.

We can think of it as conceptual knowledge, we can think of it as semantic knowledge, and we can think of it as technical skills. These ideas were introduced by Christine Borgman in 1996 and they still hold true today. Let’s define each.

Conceptual knowledge is your understanding of an information retrieval process – translating your need into a query and eventually implementing into a system.

Semantic knowledge is your understanding about how to implement a query into a specific system – or how do you input your query into a given system. And how and when to use different system features.

For example, technical skills in inputting the query: your basic computing skills, the syntax that you need in order to actually enter it into the search box. Now, today we have it a little easier; we don’t have to use any command line searching. We don’t have to put into our OPACS anymore “A:\” (A colon backslash) and then the author’s last name. And in most system’s we don’t even need to know if we need to enter the last name, first name anymore. We’ve gotten very spoiled in that way. But there are still a lot of systems out there that you really have to know the syntax in order to enter a query to make the system understand it.

2.1 Conceptual Knowledge

• Going from the vaguely-felt need…
  • To the query formulation
    • verbally expressed
    • written or typed
• Query is formulated in terms of the… (Buckland, 1991)
  • Content (words, numbers, symbols)
  • Information Entities (documents, objects)
  • “Information as Thing” (container of information)
• Successful query = translation of intended meaning into search terms

Let’s look at conceptual knowledge. Going from a vaguely felt need, where we start to realize that we need some information about this particular topic and “how can I find that information” to the idea of expressing that query, either verbally where you go and ask somebody.

You ask a reference librarian, call someone on the phone, text somebody on your phone, or type it into a search engine, type it into a database, type it into an OPAC. We then have to understand how to formulate that query within our system. We have to be able to take our actual query and put it into terms that would be useful and meaningful to either a human or a computerized system.

And oftentimes, I’m sure as you’ll learn in 5053, your user doesn’t really know what they want when they come to talk to you. When they ask you a question at a reference desk, they’ll give you a really big , broad question, when they really need this other subset of that actual query. After many years at a reference desk, I can tell you, you get everything from A - Z.

And oftentimes, you start very broadly and eventually you drill down to what they need. But again, a user often has to go through that same process when they’re seeking information.

2.2 Semantic Knowledge

• How to Implement a Query in the System
• How the System is Structured (file system used)
• Action (find, select, scan, browse)
• Access Points or Field Tags (author, title, subject)
• Search Terms (search functions of systesm: Boolean, Keyword)
• How the Search Results are Displayed

Semantic knowledge is again how to implement a query into a system. You have to know how that system is structured - what kinds of fields are in that system, what fields are searchable within a system, and the kinds of information that they would hold within those fields.

You have to know the different actions that are possible such as which fields are searchable and how to limit your query to specific field(s) in the system you are using. Other aspects include, for example, how we enter our search terms. What terms work, and as I said earlier, what kind of functions are possible in the system? Do you have to use an AND or a NOT Boolean string in order to make your search work, or can you just do fielded searching? Or if you’re in a search engine, do you just put a couple of terms into the magical box and see what you get?

You also have to know how the search results are displayed, and it’s not as difficult in some of the systems today, but remember in older systems, we had all those function keys across the top? And if you hit F1, you could print. If you hit F3, you could get a full - screen preview. Things like that. Our users again are a little more spoiled than when we were learning different systems, but still, they have to understand the different functionalities of a system. And if the system isn’t well - designed - if the navigation is difficult, or there are a lot of necessary or unnecessary features that are not easily accessible - then our users are also not going to be able to use our systems.

2.3 Technical Skills

• We MUST not take for granted that all users know how to use computers
• How the computer operates
• Syntax of entering search commands
• Commands to manipulate system (Borgman, 1996)
  • control search
  • display formats
  • control output of results

Technical skills - is the second aspect of System Knowledge.

We cannot take for granted that all our users know how to use computers. Even in today’s very highly wired web environment. Not all our users are online. And not all of our users are proficient searchers. We, then, can’t take for granted some of the basic functions of a computer, such as how a keyboard works.

Have you all heard the tech support story about the person that calls up and says, “Well, it says to hit the ‘any key,’ but I can’t find it on my keyboard. Where’s ‘any key’?” We still get stories like that. Or the one where the people were so happy because their new computer came with a cup holder - the CD drive? Thought it was for holding cups? Or even the one about the mouse, where they tried to access their computer by pointing their mouse at the screen? Or putting it down on the floor like a presser foot for a sewing machine?

These are true stories, and we still get them.

So, we can’t make assumptions that everyone understands how to use our systems.

Users base their use of systems on previous experience they have had. For example, an older user may be familiar with a presser foot when she used a sewing machine so her mental model of how to use the mouse may be based on this prior experience.

And as you all know, when you’re learing new databases, you first have to navigate a little bit. You have to look and see which kinds of fields are searchable, if it includes Boolean searches or not, if there’s any kind of proximity operators you can use to make your queries more precise.

Everyone has to go through the same kind of process whenever you start using a new system. So, they may not understand how the computer itself works, they may not know how to enter the query, and they may not know the different commands to make the system do what they need.

And, honestly, even though these are really interesting , fun stories, we also need to remember that our users don’t have the background that we all do. They’re not all from a library science program, where this is what we learn. This is what we’re here for, is to learn how to connect our users with information.

3. Task Knowledge

• Understanding of Information Problem and Use
• Affects Motivation and Goals

The third type of knowledge is what we call “task knowledge,” and this is one of the primary areas after domain knowledge that’s really essential to know about your users.

Task knowledge is an understanding a person has of their information problem and use - how well your users understand what they need, what they’re looking for and how well they understand the kind of resources they might need to use to find that information.

So, if we say that a user has high task knowledge, that means that they understand very well what the query is that they have.

They want to wallpaper their kitchen, so they know that they want some kind of book that tells them about home decoration. That’s a really pretty good understanding of their task.

But say that they need to know about the ordinances in your city if they want to build a house with a pitched roof. Where do you go for that kind of information? To the city ordinances, or your planning department, but this is not the kind of information every user will know.

But your users have to have a good understanding of the task itself and the kind of resources that they need to answer that query. And they also need to know how well your collection can help them answer that query. And this really affects our user’s motivation and goal. If they can’t easily figure out what they need, oftentimes, they will not go any further. They will leave the library or library website unsatisfied and go to another source, such as a search engine.

4. Domain Knowledge

• Understanding of Topic and Topic Organization
• Affects Ability to Exploit Language and Association

The last area, and this is probably the most important, of the four types of knowledge, is called “domain knowledge.” This is our user’s understanding of the topic and the topic organization or the discipline of the topic and how that discipline is structured. So, for example, if you’re an undergraduate chemistry student, how well at that point do you understand the discipline of chemistry? Probably you have a fairly basic understanding, but not at the same level as your professor or not at the same level as graduate students in the discipline.

This type of knowledge is important for us to know because if a user has a low or a high domain knowledge, it gives us an understanding of their ability or facility to use the language of that discipline. And when we’re creating representations, we would make a decision to use very specific terminology or to use more general terminology based upon what we understand about their domain knowledge.

So, for example, if we’re working with, again, a chemistry database, we’re assuming that our users are individuals that are in a program of chemistry - they’re either students or faculty - and we have varying levels of domain knowledge. The students probably have a low to moderate level of domain knowledge, whereas the faculty would probably have a higher level.

Example

Collection: Law Firm Library

Demographics: First Yer Law Clerks

Information-Seeking Situations: State Statute on Environmental Easements

Types and Levels of Knowledge: World, System, Task, Domain

Here’s an example on how we can apply what we’ve just been talking about.

Our collection is a law firm library. The demographics, we have first year law clerks, so they’ve probably been to law school for one year. They’re beginning to know the language of their discipline; they’re beginning to know how to conduct research in the areas. They’ve probably taken at least one legal research class. They’re beginning to understand how to find information for their particular information needs.

Our situation here is that they need to find information on a state statute on what are called environmental easements, which is a law that tells you, for example, can you have a pitched roof, or how close can you build to the water on a beach. An easement is a stipulation of ownership and ownership building rights in a particular tract of land, whether you own it, or if the government owns it.

So, this is our situation. We have the law clerks; they probably have a pretty good idea of how to go about finding out this information. They probably have a probably moderate domain knowledge - moderate to low. Their system knowledge will vary depending upon the systems they have used. And their task knowledge is probably moderate to low at this point because they may not even know what an environmental easement is at that stage in their education, but again it may vary as well.

Now, if we take the same situation and we put it in a public library , our users are going to change. We’re probably not going to have law clerks; we’re going to have a person who wants to build on the beach. And they need to know how close they’re allowed to build close to the dunes, for example. They’re probably not going to have a clue what an environmental easement is or even what a state statute is - very possible. And so their task knowledge is going to be lower, their domain knowledge is going to be lower, system knowledge depends on their experience, but all of those are going to affect how they go about looking for information, the questions they ask you, the resources they access and how they interact with different systems.

Okay, so can you see how those types and levels of knowledge and demographics can change what we need to provide in terms of access?

In a library - a public library - our records- our representations - are probably going to be at a very general level. We’re going to have one card in our catalog that represents our set of state statues if we’re fortunate enough to have them in a public library, which will have very general subject headings. Now, in a law firm library, we’re probably going to have a breakdown of the different volumes of the state statute set that tells us what’s in each individual volume. But the language in our law firm collection is going to be at a higher level of what we call of specificity - more specific terminology in use - than what we might see in a public library.

So, the choices we make when we’re designing and creating representations will be different depending upon the environment, depending upon the users, and their four types and levels of knowledge. That’s why types and levels of knowledge are so important to your understanding of organizing information.

How Do User Seek and Search for Information?

Users are motivated to seek information in a given situation to:

• answer a question

• solve a problem

• complete a task

• learn about a subject

• verify a fact

Alright, now let’s talk about our next framing question: How do users seek and search for information?

As I’ve mentioned, there are lots of different models that we’ve developed, as we’ve looked at information seeking from lots of different perspectives.

Users are motivated for many different reasons to seek information:

• to answer a question;
• to solve a problem;
• to complete a task;
• to learn about a subject;
• to verify a fact; and

many others I’m sure that you can think of!

How Do Users Seek & Search for Information?

Two Situational Levels

1. Information Seeking

   Broad context involving any or all systems or sources accessed

2. Information Searching

   Narrow context involving interaction with specific system or source

We can look at this also from two situational levels: (i) information seeking and (ii) information searching. And oftentimes in the literature, these two get used interchangeably, but they’re really not the same.

Information seeking is kind of a broader context where we look at an individual’s entire information seeking activities - from all the different systems that they use, from the beginning of their search to the end of their search and the use of the information. That would be seeking.

Searching is a more narrow context where we look at one specific interaction with either a human system, a computerized system, any of those other forms of systems that we discussed earlier. So, information seeking is a broader context; information search is where we’re looking at a specific narrow situation.

What is Information Seeking?

It refers to the user’s process for obtaining information, from recognition of an information need through use of information to resolve the information need

So, what is information seeking? Let’s go back to that idea.

The information seeking is the whole process of user recognizing their information need – that they have a gap in their knowledge, to obtaining the information itself – that they need to know something about a particular topic through the use of different systems.

Information Seeking: Assumptions

Common assumptions about information seeking in LIS are that the process is:

1. Purposeful
2. Cognitive
3. Contextual
4. Value-Based
5. Dynamic

Now in our field, as well as other fields, we’ve made some different assumptions about information seeking and the process itself - that it’s purposeful (that we’re always looking for information for a purpose).

Now, we’re changing our idea about this because of the web. A lot of people just get online and surf around. And actually, we’ve always been not just for a specific purpose when we look for information. You go to the library and you just browse the ‘new book’ shelf. You want to see what’s been published. You’re not necessarily looking for anything in particular . Maybe the purpose is to find something to read, but it’s not necessarily linked to a specific information need.

We’ve also assumed that it’s a cognitive activity - that we’re always learning as we encounter information.

That it’s contextual and that your need will change depending on the different context.

And actually, that it’s very dynamic. That as you learn more about a particular subject when you’re searching, your need will change because you might because more focused, or you might become even broader. You might go to a particular library OPAC looking for a particular subject and then you find other subjects that are really interesting and you go off on more of a non - focused track. Or you find a sub - piece of that topic, and that is really where you want to focus. So, your need is changing as well as you go into the information seeking process.

And context is important because that will impact the way that you make decisions.

For example, if you need to do some research for your research paper and it’s due in three months, right now you’ll use the web, you’ll maybe use an OPAC. You might even interlibrary loan something - a novel idea. You can still do that. But say you have three days to get your paper in, you’re going to go to the web, you’re going to access anything you can quickly access, you’re not going to interlibrary loan - you don’t have time for it now - and you’re probably going to take anything that is related to the topic that you’re working on.

Your idea of what’s relevant or useful to you changes because of the context!

Information Seeking & Searching: Models

• Motivation: why users seek information, their information needs
  • Dervin’s Sense-Making
  • Belkin’s Anomalous State of Knowledge
• Situation: how users seek information
  • Kuhlthau’s Information Search Process
• Resources: how users use resources/tools
  • Mann’s Library Research Models
• Interaction: how users interact with information systems and the process(es) they use

We’ve also framed information seeking through four basic different frameworks: motivation, situation, resources, and interaction.

For example, motivational models look at why users seek information. And what are their information needs. And how does this make each information seeking situation different? The two most prominent models in this area are Dervin’s sense - making model and Belkin’s anomalous state of knowledge (ASK).

Dervin’s sense - making model is one that’s primarily a constructivist point of view, where, and also it’s been referred to as kind of a ‘Star Wars point of view,’ in that she says that as we move through the world in our daily lives and we encounter different stimulus from different environments, we have a cognitive shift where we learn something new about the particular topic or our world around us.

And when we’re seeking information, we then have to be able to make sense of that information and fit it into our cognitive frameworks. And once we can do that, then it becomes useful to us. But again, when is that point? And what do we have to do to get there? Her model’s a lot more complex than that, and it’s really a very interesting model to study. But again, it’s more of a constructivist point of view, where we learn from and construct new realities for ourselves based upon the information we encounter in our daily lives.

Belkin’s anomalous state of knowledge, or the ASK model, is where we perceive some ‘gap’ in our knowledge - base and we look for information, we use different systems in order to fill that gap. It’s a little bit more straightforward, but it’s more of a process model from, again, where we recognize we have a gap and everything we do in the meantime is to find the information. Now, Belkin and Dervin have studied various different populations, primarily adults, and their models, again, reveal different patterns and information to us about how users look for information, the different kinds of sources they use, and how they interact with different systems.

Situational models look at how users seek information. What do they actually do ? Kuhlthau’s model, one of the most well known situational models, has affective variables to it as well. Her research has primarily been with middle and high school students.

We also have resource models. Thomas Mann who was the Librarian of the Library of Congress developed these models through watching people over the years and interacting with different people.

We have also interaction models - how users interact with information systems. For example, “how we use a particular database, how we use the web, or what systems do we actually use when we go to look for information.”

It is important to remember that each of these different frameworks and models has been developed through the years with different groups of users - from adults to children to elderly, to what we call ‘naive users’ to more experienced users, to web users, to database users, to OPAC users.

We’ve put lots of different labels on our user - groups, but we’ve developed these models on about forty years of research now in various different aspects of our field.

Bates’ Berrypicking Model

Let’s begin by looking a bit closer at a few information seeking models. One situation model that encapsulates the entire information seeking process is Marcia Bates’ Berrypicking model which depicts how users’ information needs and their understanding about that need changes as they undergo the information seeking process. The user may begin with a very general understanding of what they need to find out and how to express their information need, but as they conduct searches, interact with the reference librarian, review search results, etc. they learn more about the topic and then can refine both the information need but also how to express it within a search system. While the interaction a user has with the information system may differ, the cognitive process of how their information need changes across time and experiences, is the same for every user. This model is also related to Belkin’s ASK model as mentioned earlier.

Kuhlthau’s ISP Model

Task Initiation: Recognize Information Need

1. Topic Selection: identify general topic
2. Prefocus Formulation: investigate information on general topic
3. Focus Formulation: decide on narrower topic
4. Information Collection: gather information on narrower topic
5. Search Closure: complete the search
6. Use of Information: making use of information

Kuhlthau’s model, for example, is probably the most well - known and the most well - used in our field. She has kind of a somewhat linear approach, where we have a topic selection at the beginning, where you identify a general topic. Remember most of her research was with middle - and high - school students. The youth are looking for information for a research project or a research paper.

The first stage is topic selection.

The second stage is called ‘prefocus formulation’ where the users do some investigation on the general topic. They start looking at databases and encyclopedias. They start looking in their card catalogs to find information in their libraries about that topic - or now online.

The third stage is what Kuhlthau calls ‘focus formulation,’ where users decide on a narrower topic from that general topic that they started from. Kuhlthau links users’ affective states to each of these different stages - affective meaning different emotional responses or feelings at the time of topic selection. At the time of topic selection, they’re a little anxious. And at the prefocus formulation they’re getting even a little bit more anxious because they’re starting to look for information, they have to identify resources, they have to use different systems. And at the point of focus formulation users are starting to level off again and feeling more comfortable.

Now, why that’s important for us is because it tells us as reference librarians where we should intervene, or what Kuhlthau calls the ‘zone of intervention.’ Should we let users continue to look around for themselves for a while and then come and questions? Or should we go and say, “Can I help you with this?” Kuhlthau says to probably wait until between stages two and three, somewhere during these stages is what she calls the ‘zone of intervention.’ During the zone of intervention, that’s when you should jump in and say, “Maybe I can help you with this different system,” or “This resource might be better for you.”

Step four is where users collect the information. They’ve decided on a narrower topic, they’ve explored some different systems, some different resources, and they start gathering some information on that narrower topic.

Step five is the search closure where users complete their search, and finally, the last step that Kuhlthaus has recently added, when users use the information.

Now, this model appears to be linear, but at any point, you can go back through the stages. If you get to stage two and you don’t like this general topic, so you might propose a whole completely new topic and start over again.

Or you might get to three and you’re not quite happy with that narrow topic, so you go back and explore the general topic some more.

So, again, it’s not necessarily a linear process. But what she’s primarily famous for, and especially in this model, is this whole process, but also the affective states of the individual and when we should actually intervene.

Mann’s Library Research Models

These are models for library searching, or how people (end-users and/or professionals) use resources and tools to find information

• Search options based on organization of information in libraries
• Importance of models in concealing/revealing search options/indexing system
• Applicability of models in any subject area, language, format

Mann has developed some really interesting models. He’s looked at different search options that we’ve developed based on how we organize information. Also, his models show the importance of our library systems in revealing or concealing different pieces of information to our users. How much can they tell about our resources by looking at a bibliographic record? How useful is that? Do they actually use our pathfinders and our finding aids?

He’s also developed some different models, like for example, a specific subject or discipline model. Mann has looked longitudinally at how different disciplines look for information. For example, Social scientists do what’s called ‘citation checking or chasing’ where you find a really good article and then you go and look at the citations in the back and then you start chasing those down. People in the hard sciences, like say chemistry and physics, they call their colleagues before they do any research in our systems. They’ll actually network before they’ll use a library database or to your library catalog.

So, through the years Mann developed these different models from observing and working with people in reference situations.

The traditional library science model is one model that Mann examined how people use different tools we’ve developed such as our classification schemes, or controlled vocabulary or subject headings within library catalogs, but also how people use bibliographies and indexes.

Research Models

Examples: Mann’s Library Research Models (just a few, there are many others!)

• Specific Subject or Discipline Model
• Traditional Library Science Model
  • Classification, controlled-vocabulary catalog, published bibliographies and indexes
• Actual-Practice Model
• Principle of Least Effort

Mann’s Actual practice model is a generic model of “What do people really do when they look for information in, say, the Library of Congress?” Do they ask reference librarians first? Do they go to the OPAC first; do they go to a finding aid? So, he’s examining a user’s process rather than lumping them into a group like we saw in some of the earlier models.

The last one isn’t Mann’s model. We attribute it to Zipf. It’s the ‘principle of least effort,’ but Mann has developed this model further. This is the idea that people will go to the least amount of effort to find what they need. How many of us are guilty of that? At some point or another we all just want the information right now, right quick, and the web makes it even easier for us to do this. We can also apply this model to the Web. It’s very interesting in early retrieval research, we knew that people would probably access the first fifty hits when they searched our OPACs, and believe it or not, that still carries into the web. Most people don’t go beyond the first fifty hits when they find results on the search page.

So, these again are just some ideas of the different models we’ve developed through the years.

Shannon & Weaver Communication Model

There are also models that help us further understand the issues that may develop as part of the interaction between users and human or computerized information retrieval systems. Shannon and Weaver’s classical model of communication depicts the process by which a sender conveys their information need or message to an information system, whether it is a human or computerized system.

This model shows this transaction between two people but in LIS we often extend it to include computerized or online systems as well. The user conveys their information need by conveying a message, which is transmitted through a channel (spoken language, or a query entered into a search engine), which is then received by the receiver or information system. Of course much can go wrong during this transaction if noise is introduced into the process. The message can be corrupted, thereby affecting how the receiver hears and interprets the message.

This noise can affect the information received/retrieved back as well. This model can be used to further understand the reference interview transaction and the possible issues when users try to convey their information need to the librarian or system.

Taylor’s Levels of Questions Formation

• Visceral Need: actual but unexpressed need that changes in form, quality, concreteness as user experiences everyday life

• Conscious Need: conscious with-in brain description of an ill-defined area - may be a rambling statement or unfocused

• Formalized Need: user is able to formalize a statement in concrete terms

• Compromised Need: representation of the user’s info need within the system

Your textbook (p. 29) also introduces Taylor’s levels of question formation that are indeed helpful in understanding the cognitive types of ways users express their information needs users during the information transaction. This typology is especially important as we discuss the reference interview, the search process, and tactics of searching. For example, the visceral need refers to the actual but unexpressed need a user may have that changes in form, quality, and concreteness and the reference interview and their seeking process proceeds. Or the compromised need (though I am not fond of this label) is how the user represents their information need within a search system. We will come back to this typology later when we learn about search strategies and the reference interview.

Reasons WHY People Don’t Reveal their Real Information Needs

• Nature of the reference interview
• Difficulty of developing a searchable topic
• Difficulty of expressing one’s info need to someone else
• Assumptions users have about libraries and their resources
• Assumptions users have about librarians
• Users want to be seen as competent, secure, and knowledgeable

Chapter 3 of your textbook (p. 36-37) introduces an intriguing dimension to understanding the reasons why users do not reveal their real information needs during the reference interview. The table on pages 36-37 of the textbook gives more detail about each but as an information professional interested in helping users negotiate their query and search systems, these reasons are quite revealing and worth reviewing.

For example, the nature of the reference interview. Users attempt to anticipate what the reference librarian will need to know to help them find information. Oftentimes the initial interview focuses on helping the user to articulate their information need. As mentioned in earlier lectures, expressing one’s information need to someone else or within an information systems’ parameters is difficult and requires knowing more about how the system works, and how your query terms may be processed by a system. Users very often have difficulty negotiating how to express their real information need and then formulating a search within the system.

It is particularly intriguing to me that users want to be seen as competent, knowledgeable searchers. They also may not want to reveal the real reason for their search or any personal information that might put their security at risk.

Users may think the librarian will consider they incompetent or that they are cheating on an assignment by asking them for help. We will come back to this discussion later when we taught about the reference interview.

Chowdhury Reminds Us

• Context of the user
• Nature of the information
• Context for the information need
  • work environment (nature of work or profession)
  • task to complete
• Demographic variables
• Social groups

Chowdhury, in Chapters 10 and 11 reminds us to keep the context in which the user exists in mind when trying to understand users and their information behaviors and systems use.

For example, the context or environment in which the user exists can tell you about the type of information need a user may have, the types of resources necessary to satisfy that need, or the specialized collection or system necessary to satisfy the information need. For example, an academic library user will bring different types and levels of knowledge to the IR transaction. Academic libraries serve students, faculty, administrators, and others. They contain collections and systems designed to serve these user groups. As a result, the type of information needs will likely be quite different than those of public library users. Also, when thinking of the context of the information need, the work environment and nature of the profession is important. If the user works for a large corporation as a researcher, or an information analyst, they will have very specialized information needs specific to that environment. Their tasks will greatly differ than those of an academic or public library user, and they will bring different domain and task knowledge to the information seeking process.

These are but a few examples to illustrate how context affects the information seeking and retrieval process. Review Chowdhury Chs. 10 and 11 for further discussion on contextual factors and additional information seeking models to consider.

Implications for Practice

Models of Information Seeking....

• Guide development of user-centered services
  • Reference
  • Searching
  • Indexing/Cataloging
  • IR Systems Design
• Librarians improve organization of resources to accommodate users’ searching styles
• Understand users’
  • Decision-making criteria
  • Process strategies
  • Expectations of search process
  • Attitudes and feelings
  • Organization schemes (categories, labels, methods)

Why is understanding the many complexities of users important to what we’re talking about in this class? Why am I boring you with all this stuff?

First, understanding our users and their information behaviors and organizing behaviors guide development of user - centered services, not just in reference departments, not when we search, help our users search, but also in indexing and cataloging.

The models we have developed and the research conducted into user behaviors help us understand more about the kinds of searches people do. What fields do they use in our databases? Do they only conduct author - title, which are called known - item searches, or do they do more subject searching?

So, when we’re building a system, we’re designing a system that we hope will be useful. When we’re creating representations, such as card catalog records, knowing about how people go about looking for information can help us design better systems. We’ll know which fields need to be searchable; we’ll know which attributes or aspects of our objects should be represented in our records.

So, again, knowing a little bit more about how our users look for, and find information, and use information, can help us design systems that are more in line with their information needs. This understanding also can help us choose in terms when indexing and cataloging and to know more about what level of terminology we need to use when representing the subject content.

Furthermore, as a searcher, understanding the levels of exhaustively and specificity that are present in the indexing or the cataloging in the system can help you become a better searcher. Your search terms will be more aligned with the actual vocabulary used by the cataloger or indexer within the system. So, therefore, you should have better retrieval. You can also help your users understand this and make them more efficient in their searching as well.

Implications for Practice (Cont.)

• Help us
  • Choose elements for system design
    • Fields to include in system
    • Fields to make searchable (indexed)
    • Creation of input rules
  • Choose appropriate controlled or other vocabulary to use (natural language, free text)
  • Choose appropriate level of indexing
    • Exhaustivity
    • Specificity
    • Number of subject terms to include

Other implications for practice that I did start to mention on the previous slide are that understanding our users really helps us to design a particular system for users. In libraries we tend to be very generic in that our systems are for a very large, diverse audience of individuals.

So, sometimes it’s not possible to build a system that’s going to satisfy every user. In fact, it’s really not possible to do so, unless you have a very specialized collection, a very specialized user - group and you understand very closely those users and their information behaviors.

But learning about our users helps us to choose elements that we want to include or that we know need to be included in our system design. It helps us make decisions and to work with our vendors to determine what fields to include and also which fields should be indexed or searchable within your system. It also helps us to create input rules.

So, for those of you who are going to work on the organization system project, there is a section in the project where you have the opportunity to create input rules for each of the fields within your system. This is analogous to the standards we use when we create catalog records in library catalogs. We use the Resource Description and Access (RDA) and other subject heading lists in order to create records in our system, and those standards are sets of rules that guide catalogers.

Also, we use what we know about users to choose appropriate controlled vocabulary or other types of vocabulary to represent subjects in our systems. In libraries we tend to use Library of Congress Subject Headings (LCSH) or Sears subject headings if we’re in a school library. We might use a user - specific ontology if we’re in a business community, where we’ve designed a specific ontology to represent the language of that community. So, again, knowing about our users and the types of searches or searches they do conduct really can help us in these choices.

Also, it helps us choose the appropriate level of indexing in our catalogs. How many subject terms do we need? How many fields do we need? And there are two measures that we talk about in libraries called ‘exhaustivity’ and ‘specificity’ that really define the number of subject terms or the level of exhaustiveness - how many terms do we use to describe the content of an object for effective retrieval? How specific should those terms be? Is it appropriate to have very specific terms in the catalog or an academic library versus when we’re creating records for children, where we’ll want to have more broad generalized terms?

So, again, learning more about our users and the types of searching systems they use really can help us determine all of these different factors.

User Expectations

• Expectations for Retrieval Performance (eg. precision and recall)

• Other Measures: Relevance or Pertinence

User expectations of the information seeking process and systems are also changing. Generally, measures of users’ expectations are accomplished by assessing precision and recall.

So, for example, our users have very specific expectations for retrieval performance. Some of the users will be very specific that they want everything on a particular topic, meaning they’re casting a very broad net. And there might be some that just want one or two items.

By examining their questions, we can get an idea of their expectations for retrieval, and we measure their expectations and how well our system has done at providing the information the user expects through measures of precision and recall. We’ll talk more about this in later module, but I wanted to address it just briefly here. Other measures that you might see in the literature related to retrieval are ‘relevance’ or ‘pertinence’–or how useful that information or that object is to the user.

Recall

Recall assesses completeness of output and ability of system to retrieve all relevant items

So, let’s talk about recall first.

Recall assesses the output of the system, meaning that every object that is related to that query in some way should be returned back to a user in the list of hits in the system. It also assesses the ability of the system to retrieve all relevant items, meaning that everything that is retrieved in the set should be related to the subject or the attributes in the user’s question.

So, if they’re looking for all books on paint treatments that you might have in your collection, they should return all records related to paint treatments. If not, it’s said to be ‘low recall.’

Precision

Precision assesses relevance of output and ability of system to retrieve only the relevant items

Precision, on the other hand, assesses relevance of output, meaning it assesses the ability of the system to retrieve only the relevant items. So, precision is looking at how specifically the system has returned only items that are appropriate to your query. As you can see, this could be a problem, especially if the system is difficult to use or the representations don’t match the terms being used by the user themselves.

And as we’ll talk about it in later module, there are very different types of systems and they operate in different ways, and that could impact both recall and precision.

So, again, when we look at our users’ queries, we can tell whether or not they want a very precise return, and only the particular objects, such as only books by Janet Evanovich or current books by a particular author, and if they retrieve items that are not books or they retrieve items not by that author, it’s said that precision is very low. It’s also important to understand that our systems have an inverse relationship between precision and recall, in that precision if it’s high, means that recall is generally pretty low.

IR System Design

“Why are online catalogs still hard to use?” (Borgman, 1996)

• System Design Models
  • Query Design
  • Online Card Catalog Model
  • Representations have to fit into “cookie cutter” systems that were developed many years ago and do not necessarily benefit from what we know about users and their informatoion seeking processes
• Often times specialized knowledge is required
  • Semantic
  • Syntactic

Christine Borgman in her 1996 article, “Why are online catalogs still hard to use?” explained why OPACs in libraries are difficult to use. Her article is a really good one for giving you even more conceptual framework for what we’ve addressed in this lecture.

System design models in the past were based upon the original card catalog model and what could fit on the library catalog card. And unfortunately, a lot of systems today are still based on that model. We can fit on information about the author, the title, and maybe one to three subject headings, and a classification code.

So, that’s how our basic library systems were designed. In our next generation catalogs we have improved the interface design, added more searching functionality, and allow users to search using a “Google - like” search box. BUT, and this is what is really important, their retrieval is still based on the RECORDS or representations of items in the collection that catalogers have developed.

IR System Design (Cont.)

Traditionally IR System Design Often...

• Based on assumptions about information-seeking rather than research (although that is changing!)
• Directed toward system use by information professionals, not end-users
• Forces system users to adapt to the system, rather than adapts to users’ natural abilities

In our field and as new technologies are developed and as users start to use different systems and start to use them in different ways, it also forces us to rethink decisions and choices we’ve made in the past. Again, that’s why I think it’s such an exciting time to be in our field right now because we are really at the forefront of a lot of these change.

Also, we’ve been designing and working through our records structures, we’ve been designing XML systems that can work XML MARC, which gives us even more functionality than the traditional MARC system might have. And people are designing more user - centered systems when they create digital libraries; we’re not just using MARC record structures, we’re designing metadata schemes like Dublin Core.

What is in our library catalog records may not have changed much until the recent 5 years with the development of the FRBR model and with the new cataloging standard, RDA. AND when we move beyond the MARC record in library catalogs into BIBFRAME, our records will hopefully connect to other external records through this linked data structure. But that is for another module.

IR System Design (Cont.)

Good IR System Design Tends To Be...

• Based on study of users needs/interactions
• Interactive to accommodate natural information-seeking behaviors
• Flexible to accommodate differences among individual users

In conclusion, good information retrieval system design should be based on a study of our users, user needs, and their interactions with different systems.

We need to understand the types of searches that they bring to a system. We need to understand the attributes in those searches that should be fields in our system. We also need to understand the exhaustivity and specificity level of the terms that we use to represent subjects or other concepts within our systems. And also, whether or not we need to have very complex concepts represented. Or can the terms be very general in nature?

Our systems also have to be very interactive to accommodate natural information seeking behaviors. We need to think about how users really do interact with systems at different levels - at their homes, in their work environments - to try to make sure our systems can try to emulate some of those natural behaviors.

Systems also need to be flexible to accommodate differences among different users because no two users are going to interact with the system in the same way. They’re not going to use the same terminology. They’re probably not going to use the same functions in your system - whether it’s a simple search or an advanced search. So, you’re going to need to keep that in mind as well.